Professional Documents
Culture Documents
The Classical Stefan Problem Basic Concepts Modelling and Analysis With Quasi Analytical Solutions and Methods Gupta S C Full Download Chapter
The Classical Stefan Problem Basic Concepts Modelling and Analysis With Quasi Analytical Solutions and Methods Gupta S C Full Download Chapter
New Edition
No part of this publication may be reproduced or transmitted in any form or by any means, electronic or
mechanical, including photocopying, recording, or any information storage and retrieval system, without
permission in writing from the publisher. Details on how to seek permission, further information about
the Publisher’s permissions policies and our arrangements with organizations such as the Copyright
Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/
permissions.
This book and the individual contributions contained in it are protected under copyright by the Publisher
(other than as may be noted herein).
Notices
Knowledge and best practice in this field are constantly changing. As new research and experience
broaden our understanding, changes in research methods, professional practices, or medical treatment
may become necessary.
Practitioners and researchers must always rely on their own experience and knowledge in evaluating and
using any information, methods, compounds, or experiments described herein. In using such information
or methods they should be mindful of their own safety and the safety of others, including parties for
whom they have a professional responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any
liability for any injury and/or damage to persons or property as a matter of products liability, negligence
or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in
the material herein.
ISBN: 978-0-444-63581-5
Some symbols and abbreviations which are common throughout the book are given below. They have
also been explained in the text, occasionally, but not every time they occur. In addition to these symbols,
several other symbols have been used in the text, and they have been explained at the places they occur.
The parameters occurring in the equations could be dimensionless quantities or may have dimensions.
At those places where the equations are in the dimensionless form, the method of dimensionalization has
been mentioned or referred to reference indicated in the text.
For the notations used for the function spaces, the reader is referred to Appendices A–D. The same
have been explained in the text, occasionally.
Standard notations have been used for the numbering of equations, figures, definitions and proposi-
tions. For example, Eq. (7.2.7) refers to the seventh equation in the second section of Chapter 7.
xv
xvi List of Symbols
Subscripts
Greek Symbols
Abbreviations
The first edition of The Classical Stefan Problem: Basic Concepts, Modelling and Analysis
with Quasi-Analytical Solutions and Methods was published by Elsevier in 2003 as a volume
in the North-Holland Applied Mathematics and Mechanics book series. The main objective
was to discuss comprehensively, insofar as possible, the theoretical aspects of classical
formulations and analysis of some of the topics of the study of the Stefan problem.
The Stefan problem, which has some characteristic features, forms only a small part of
a bigger class of problems known as Free Boundary Problems. Even in 2003, the existing
literature on Stefan problems was so vast that it seemed feasible to discuss only classical
formulations of Stefan problems related to topics such as supercooling, variational inequality,
hyperbolic Stefan problems, inverse problems, existence and uniqueness and other aspects of
analysis. While discussing variational inequalities, inverse problems, analysis aspects, etc.,
the discussion of weak solutions was unavoidable, and they were discussed as needed and
not comprehensively. There are weak solutions, which are as good as classical solutions. To
bridge the gap between other basic sciences and mathematics, and to deepen the understanding
of the book’s contents, some definitions, theory and results from thermodynamics, metallurgy,
physics, applied mathematics, etc., were included as separate chapters.
A 5-year effort by the sole author produced the earlier edition. The positive reviews and
readers’ indirect encouragement provided the author inspiration and courage to undertake
this new project. This new edition features an extensive Chapter 12, which deals with
quasianalytical solutions and methods of classical Stefan and Stefan-like problems. Because
the class of Stefan-like problems is very large, only those problems whose formulations
are similar to those of Stefan problems and whose physics and formulations can be easily
explained have been included. Rather than publish the contents of Chapter 12 as a separate
book, with the aim of bridging theoretical and solution aspects of Stefan problems, Chapter 12
has been introduced along with earlier chapters.
A solution method is a procedure, and it is not confined only to Stefan problems. It
can be applied to any mathematical physics problem. However, for illustrative purposes, the
discussion is focused only on formulations of classical Stefan problems and some Stefan-like
problems. It is easier to explain the method with clarity in a concise way than the solution, as
describing the solutions requires too much space.
Chapter 12 is divided into 10 sections and each section into several subsections. First,
Section 12.1 begins with an overview of the aims, objectives, and contents of the chapter.
Some preliminaries, such as Green’s functions in various geometries, similarity variable, and
similarity solution, are discussed.
A lengthy Section 12.2 is devoted to exact analytical solutions pertaining to various ge-
ometries, including ellipsoidal and paraboloidal. Sections are assigned to various geometries,
different types of heat equations (such as with parameters depending on temperature and
xix
xx Preface to the New Edition
space variables), Stefan problems with kinetic conditions, equations with fractional derivatives,
multiple-phase problems and dilute binary alloy problems.
Section 12.3 is about series solutions of various types, including short-time solutions.
Section 12.4 deals with the analytical-numerical solutions of Stefan problems. Here the
term analytical-numerical is used for those solutions in which a complete analytical solution
cannot be obtained, and after some analytical derivation, numerical solutions are attempted
with the help of some suitable numerical schemes. The analytical derivation part should be
dominant and should have some variety if possible. Terms, such as semianalytical solution
and quasianalytical solution, are also used, but we prefer analytical-numerical solution. The
Adomian decomposition method, variational iteration method, integral equation approach and
regularization of Dirac-delta function are also discussed.
Section 12.5 is about analytical-numerical solutions of inverse Stefan problems. In
addition to the methods discussed in Section 12.4, the homotopy analysis method and some
regularization methods are also discussed.
The analytical-numerical solutions of hyperbolic Stefan problem are discussed in Sec-
tion 12.6, with the background information provided in Chapter 8. A rigorous background of
deriving Green’s function in the planar case is also briefly discussed.
Section 12.7 is about the use of complex variable methods in solving solidification/melting
problems and Hele-Shaw problems. The singularity development in suction problems, types
of singularity and its possible removal are described.
Approximate solutions and methods are discussed in Section 12.8. A major portion
of this section is devoted to the heat-balance integral method and its refinements and
variations, such as RIM, ARIM and hybrid methods. Weighted residual methods, such as the
Galerkin method and the orthogonal collocation method, are discussed briefly with only a few
illustrative examples. This section also discusses the first variation, variational principles and
the derivation of Euler’s equation for a given functional using calculus of variations. Finally,
the section describes the method of constructing a functional for a given problem whose first
variation or Euler’s equation will be the required differential equation.
A considerable amount of literature exists on perturbation solutions because of their
easiness in application. Therefore the emphasis in Section 12.9 is on applications of the
homotopy analysis method and the homotopy perturbation method. Regular perturbation and
singular perturbation methods applied to solutions of Stefan problems in various geometries
as well as a variety of formulations are discussed. Applications of the methods of strained
coordinates and matched asymptotic expansions are also illustrated.
Section 12.10 offers brief reviews of some supplementary references. Chapter 12 con-
cludes with an extensive bibliography of about 455 references.
The presentation of material in all 12 chapters is characterized by discussions based on the
thorough study of full-length research papers. The discussion includes my own comments
on many published works in Chapter 12. Reporting purely numerical solutions was never
the objective of the chapter, as that would require several separate volumes. However, for
analytical-numerical solutions, the highlights of numerical solutions and results are given very
briefly along with names of the software used if given in the referenced paper.
Invariably, the author thinks conceptually in terms of obtaining the solution first and then
devising a method to achieve it. This is why solution comes first in the book title, followed by
method.
Preface
This volume emphasizes studies related to classical Stefan problems. The term ‘Stefan
problem’ is generally used for heat transfer problems with phase-changes such as from the
liquid to the solid. Stefan problems have some characteristics that are typical of them, but
certain problems arising in fields such as mathematical physics and engineering also exhibit
characteristics similar to them. The term ‘classical’ distinguishes the formulation of these
problems from their weak formulation, in which the solution need not possess classical
derivatives. Under suitable assumptions, a weak solution could be as good as a classical
solution. In hyperbolic Stefan problems, the characteristic features of Stefan problems are
present but unlike in Stefan problems, discontinuous solutions are allowed because of the
hyperbolic nature of the heat equation. The numerical solutions of inverse Stefan problems,
and the analysis of direct Stefan problems are so integrated that it is difficult to discuss one
without referring to the other. So no strict line of demarcation can be identified between a
classical Stefan problem and other similar problems. On the other hand, including every related
problem in the domain of classical Stefan problem would require several volumes for their
description. A suitable compromise has to be made.
The basic concepts, modelling, and analysis of the classical Stefan problems have been
extensively investigated and there seems to be a need to report the results at one place. This
book attempts to answer that need. Within the framework of the classical Stefan problem
with the emphasis on the basic concepts, modelling and analysis, I have tried to include
some weak solutions and analytical and numerical solutions also. The main considerations
behind this are the continuity and the clarity of exposition. For example, the description of
some phase-field models in Chapter 4 arose out of this need for a smooth transition between
topics. In the mathematical formulation of Stefan problems, the curvature effects and the
kinetic condition are incorporated with the help of the modified Gibbs–Thomson relation.
On the basis of some thermodynamical and metallurgical considerations, the modified Gibbs–
Thomson relation can be derived, as has been done in the text, but the rigorous mathematical
justification comes from the fact that this relation can be obtained by taking appropriate limits
of phase-field models. Because of the unacceptability of some phase-field models due to their
so-called thermodynamical inconsistency, some consistent models have also been described.
This completes the discussion of phase-field models in the present context.
Making this volume self-contained would require reporting and deriving several results
from tensor analysis, differential geometry, nonequilibrium thermodynamics, physics and
functional analysis. I have chosen to enrich the text with appropriate references so as not to
enlarge the scope of the book. The proofs of propositions and theorems are often lengthy and
different from one another. Presenting them in a condensed way may not be of much help to
the reader. Therefore only the main features of proofs and a few results have been presented to
suggest the essential flavour of the theme of investigation. However at each place, appropriate
references have been cited so that inquisitive readers can follow them on their own.
xxi
xxii Preface
Each chapter begins with basic concepts, objectives and the directions in which the subject
matter has grown. This is followed by reviews—in some cases quite detailed—of published
works. In a work of this type, the author has to make a suitable compromise between length
restrictions and understandability. I have followed my best judgement in this regard. I hope the
readers will appreciate my efforts.
S.C. Gupta
Bangalore
Acknowledgements
I gratefully acknowledge the financial support from the Department of Science and Tech-
nology, Ministry of Science and Technology, Government of India, without which it would
not have been possible for me to undertake this book-writing project. The administration
of the Indian Institute of Science, Bangalore, and the Department of Mathematics, provided
infrastructural support for which I am extremely thankful. During my visit to the University of
Oulu, Finland, in 1997, I collected lot of material relevant for this book. I sincerely thank my
host, Professor Erkki Laitinen and Department of Mathematics, Oulu, for their kind hospitality.
The idea of writing this book nucleated during my visit to University of Rosario, Argentina, in
1991, and fructified 11 years later. I thank Professor D.A. Tarzia, Department of Mathematics,
Universidad Austral, Rosario, for sending me the bibliography prepared by him, and some
reprints which I could not get from other sources.
I am grateful to Professor Adimurthi, Department of Mathematics, TIFR, at IISc,
Bangalore, and Professor A.K. Lahiri, Department of Metallurgy, Indian Institute of Science,
Bangalore, for reading and responding to some portions of this book. My grateful thanks
to Dr. Thomas Chacko, Professor of English and Communication Skills, Department of
Management Studies, for style editing this volume with a great deal of attention. There are
many old friends and well-wishers who sent reprints and preprints that proved very useful for
me. Without naming them individually, I thank all of them for their selfless help.
I thank the Publishing Editor and staff of Elsevier and the Series Editor Professor
J.D. Achenbach of the book series ‘Applied Mathematics and Mechanics’ for their cooperation.
The excellent typing in Latex and drawing of figures in computer graphics were done by
Susheel Graphics, Bangalore. Without their cooperation the work would not have taken the
final shape.
I dedicate this New Edition to my wife whose constant inspiration and support helped
me in completing this long-term project. In the initial stages the help rendered by Dr Bryan
Davis, Solutions Production Director, Science and Technology Series, Elsevier and at the later
stage the constant cooperation given by the Editor, Dr A. Koch and the Project in charge,
Ms Amy Clark is gratefully acknowledged. The offer from the Chairman Department of
Mathematics, Indian Institute of Science, Bangalore, to use some of available facilities of
the Department and some contingency grant from the Centre of Continuing Education, Indian
Institute of Science, Bangalore, is an unforgettable experience for a retired Professor. Without
naming any individual, I thank the authors who sent me reprints. The production of the
manuscript was well handled by S&T book production centre at Chennai, Tamil Nadu, India.
xxiii
Chapter 1
The term Stefan problem can be best understood with the help of an example for which the
reader is referred to Section 1.3. Our concern in the present section is to understand the notion
of a free boundary which is a typical feature of the Stefan and Stefan-like problems. Therefore
this chapter begins with some examples of Stefan and Stefan-like problems which demonstrate
the existence of an unknown boundary, commonly known in the literature as a ‘free boundary’
or a moving boundary. In the context of solidification/melting problems, with which Stefan
problems are commonly associated, the free boundary is also called a phase-change boundary
or a melting front or a freezing front. Some authors use the term free boundary when the
unknown boundary is static and moving boundary when it is time dependent. In this volume
we do not make any distinction between ‘free’ and ‘moving’ boundaries. The term boundary
is used for a surface also. In many of the examples considered in this section, the identification
of the free boundary and the mathematical formulation of the problem are rather easy but in
some of them even the identification of the free boundary is difficult. The problems given here
are from various fields of mathematics, physics and engineering and demonstrate the existence
of free boundaries. Our interest in examples given in this section is more on demonstrating the
existence of a free boundary and its typical characteristics than justifying the formulation.
Problem 1.1.1 (Steady-State Heat Conduction With a Free Boundary). Find the steady-
state temperature T(x, y) satisfying the equation
∂ 2T ∂ 2T
+ = 0, (1.1.1)
∂x2 ∂y2
If f (x, y) is known throughout the plane, then one boundary conditions on R1 will be
∇ (T − f )|R1 = 0. (1.1.4)
The problem is to determine the temperature T(x, y) in D, and the unknown boundary R1 .
Two boundary conditions are to be prescribed on R1 . One, because it is a boundary and one
more boundary condition is required to determine an unknown boundary. Note that R1 can be
determined only by solving the system (1.1.1)–(1.1.4). Such an unknown boundary is called a
free boundary or a moving boundary. In ‘boundary value problems’ of mathematical physics,
the boundary of the region under consideration is completely known. Conditions (1.1.3)
and (1.1.4) are called free boundary conditions and (1.1.2) is a fixed boundary condition.
Although Eq. (1.1.1) is linear, free boundary problems are nonlinear problems because of
the nonlinearity of the boundary conditions at the free boundary.
Problem 1.1.2 (Steady-State Free Surface Flow With Surface Tension). Consider a two-
dimensional steady, incompressible, irrotational flow in a long channel. x and y axes are taken
along the length and depth of the channel, respectively, with the bottom of the channel taken as
y = 0 and the upper surface of water, as a free surface or a free boundary denoted by y = η(x),
where η(x) is unknown. If u(x, y) and v(x, y) are the velocity components in x and y directions,
respectively, then
∂u ∂v
+ =0 (equation of continuity), (1.1.5)
∂x ∂y
∂v ∂u
− =0 (irrotationality condition). (1.1.6)
∂x ∂y
where H is the given total water head of water, ρ is the density, g the acceleration due to gravity
and q is the velocity vector. If the bottom of the channel is a rigid boundary, then
v = 0, on y = 0. (1.1.8)
and
q · n = 0. (1.1.10)
Here, p̄ is the atmospheric pressure (known), σ is the surface tension (known) and Rc is
the radius of curvature of the free surface, taken as positive when the centre of curvature
is situated above the free surface and n is the unit outward normal to the free surface. Rc
can be expressed in terms of derivatives of η(x). Free boundary has been taken as static in
The Stefan Problem and Its Classical Formulation Chapter | 1 3
this problem. Eq. (1.1.10) implies that the normal component of the fluid velocity at the free
boundary is zero.
Problem 1.1.3 (Free Surface Flow With Time Dependent Free Surface). Let the free
surface in Problem 1.1.2 be time dependent and represented by y = η(x, t) with η|t=0 being
given. Surface tension effects will be neglected. If the velocity field q is expressed as
q = ∇φ, (1.1.11)
then from Eqs (1.1.5), (1.1.6) it is easy to conclude that φ(x, t) satisfies the equation
∂ 2φ ∂ 2φ
+ = 0, b(x) < y < η(x, t). (1.1.12)
∂x2 ∂y2
Here, y = b(x) is the equation of the bottom of the channel. The momentum equation can be
written as (cf. [3])
∂ q 1
− grad(p/ρ),
+ grad |q|2 − q ∧ curl q = F (1.1.13)
∂t 2
represents body forces. On substituting q from Eq. (1.1.11) in Eq. (1.1.13) and integrating
F
with respect to x, we obtain
∂φ 1
+ |∇φ|2 + gy + p/ρ = ψ(t), (1.1.14)
∂t 2
provided the density is taken as constant and the gravitational field is the only force field. The
arbitrary function ψ(t) can be absorbed in φ(t) and Eq. (1.1.14) becomes
∂φ 1
+ |∇φ|2 + gy + p/ρ = 0. (1.1.15)
∂t 2
∂φ
= 0 on y = b(x), (1.1.16)
∂n
where n denotes the unit outward normal to y = b(x). On the unknown free boundary
y = η(x, t), the two conditions are given by
∂φ 1
+ |∇φ|2 + gη(x, t) + p/ρ = 0, (1.1.17)
∂t 2
and
∂φ
= V · n, (1.1.18)
∂n
where V is the velocity of the free boundary and n is the unit outward drawn normal on it.
Eq. (1.1.18) can be expressed in terms of quantities already defined. Let,
Then
∂F ∂F ∂F
DF = dx + dy + dt = 0, (1.1.20)
∂x ∂y ∂t
or
∂F ∂η dx dy
− =− + = V · ∇F = (V · n)|∇F|, n = ∇F/|∇F|. (1.1.21)
∂t ∂x dt dt
or
∂η ∂φ ∂η ∂φ
=− + . (1.1.23)
∂t ∂x ∂x ∂y
Eq. (1.1.12) is to be solved using the boundary conditions (1.1.17) and (1.1.23), the fixed
boundary condition (1.1.16) and the prescribed η(x, t) at t = 0. In this problem the velocity
of the fluid is not time dependent but the free boundary is time dependent. Such problems
are called quasi-steady state free boundary problems or degenerate free boundary problems.
We shall see later that the term degenerate free boundary problem is used for other types of
problems also.
εf (x, t) can be considered as the disturbance at the flat surface y = y0 . Substituting Eq. (1.1.24)
in Eqs (1.1.12), (1.1.16), (1.1.17), (1.1.23), and in the changed equations retaining only linear
terms in ε and dropping higher order terms of ε, we obtain
∂u
= 0 on y = b(x), (1.1.26)
∂n
∂u
+ gf = 0 on y = y0 , (1.1.27)
∂t
∂u ∂f
= on y = y0 . (1.1.28)
∂y ∂t
In Eq. (1.1.27), we have taken p = −ρgy0 which comes from the contribution of the zeroth-
order terms in Eq. (1.1.17).
The Stefan Problem and Its Classical Formulation Chapter | 1 5
∂ 2u ∂u
+g = 0 on y = y0 . (1.1.29)
∂t2 ∂y
Since y = y0 is not a free boundary, only one boundary condition is to be prescribed on it. For
further information about Problems 1.1.2 and 1.1.3, see [3–5].
Problem 1.1.4 (A Problem of Reproductive Toxic Mass Diffusion). Let u(x, t) be the
concentration of a toxic mass which is diffusing in a region , where
= {x ∈ R : 0 ≤ x ≤ 1}. (1.1.30)
If the concentration exceeds a certain value uv in a portion of , then it is called a toxic region.
Let the reproduction rate of toxic mass in the toxic region be P and in the nontoxic region αP,
0 < α < 1. The toxic and nontoxic regions are separated by a surface S, where
and
u(x, t) and the free boundary x = φ(t) are to be obtained by solving the following system of
equations. For u > uv
For u < uv
The term d0 + d1 u accounts for the mass loss due to the bottom leakage, and other similar
factors. d0 , d1 , α and P are positive constants. Mass diffusion coefficient has been taken to be
unity which is possible by suitably defining the time and/or length scales.
and
u(φ(t), t) = uv . (1.1.36)
It may be noted that the velocity of the free boundary is not explicitly occurring in Eq. (1.1.35)
which was so in Problems 1.1.2 and 1.1.3. The boundary conditions at the free boundary
in which the velocity of the free boundary is not occurring explicitly are known as implicit
free boundary conditions. Eqs (1.1.33)–(1.1.36) are to be supplemented with a suitable initial
condition at t = 0 and with boundary conditions at the fixed boundaries x = 0 and x = 1.
Some results on the existence of solution of the above problem and the regularity of the free
boundary can be found in [6].
6 The Classical Stefan Problem
Problem 1.1.5 (Gas Flow Through Porous Media). The equation of state for an isentropic
(constant entropy) flow of an ideal gas in a homogeneous porous media is given by [7]
ρ(x, y) = ρ0 pα ≥ 0, (1.1.37)
where ρ(x, y) is the density and p(x, y) is the pressure. ρ0 ∈ R+ and α ∈ (0, 1] are constant.
The conservation of mass gives
∂ρ
= −γ
div(ρ V) , (1.1.38)
∂t
where γ is the porosity of the medium. According to Darcy’s law [8], the velocity V of the gas
flow in a porous medium is given by
β ∈ R+ is the permeability of the medium and η ∈ R+ is the viscosity of the gas. V and p can
be eliminated from Eqs (1.1.38), (1.1.39) and we obtain
∂ρ β
= 1/α
∇ 2 (ρ m ), ρ ≥ 0, (1.1.40)
∂t ηγρ0 (1 + α)
where m = 1 + 1/α. The diffusion is called ‘fast’ if m > 1, and ‘slow’ if m < 1.
By suitably choosing the time and/or length scales, the following equation can be obtained
from Eq. (1.1.40):
∂ρ
= ∇ 2 (ρ m ), ρ ≥ 0. (1.1.41)
∂t
If α ∈ (0, 1], then m ∈ [2, ∞). Eq. (1.1.41) is the porous media equation and it arises also in
other contexts such as population dynamics and plasma physics [7]. In order to calculate the
mass flux of the gas, the r.h.s. of Eq. (1.1.41) can be written as
mρ m−1 grad ρ is the mass flux and mρ m−1 can be taken as diffusivity. Diffusivity vanishes
as the density ρ tends to zero. Therefore Eq. (1.1.41) is a nonlinear degenerate equation in
the neighbourhood of any point where ρ = 0 but is nondegenerate and uniformly parabolic
(see [9] and Eq. (7.3.26)) in the neighbourhood of any point at which ρ is away from zero.
Such problems are called degenerate parabolic-elliptic problems. An important consequence
of nonlinear degeneracy is that there is a finite speed of propagation of a disturbance from rest
which is in contrast to the parabolic heat equation in which the speed of heat propagation is
infinite. The finite speed of propagation may give rise to waiting time solutions. Eq. (1.1.41)
is to be supplemented with an initial condition if the region is infinite and with both initial and
boundary conditions if the region considered is finite. The existence of a free boundary in such
problems can be illustrated with the help of the following example. Let
∂ρ
= ∇ 2 (ρ m ), −∞ < x < ∞, t > 0, (1.1.43)
∂t
The Stefan Problem and Its Classical Formulation Chapter | 1 7
> 0 for x ∈ RI = (a1 , a2 ), −∞ < a1 < a2 < ∞
ρ(x, 0) = (1.1.44)
= 0 for x ∈ R\RI .
This problem, generally, does not have a classical solution. The classical solution of a
problem can be roughly stated to be a solution in which the dependent variable possesses
continuous derivatives of the order required in the problem formulation. The mathematical
definition of a classical solution will be discussed later but at present it would suffice to state
that the solution ρ(x, t) of Eqs (1.1.43), (1.1.44) may not possess the required continuous
derivatives. For t > 0, gas will be diffusing to the right of x = a2 and to the left of x = a1 and
thus giving rise to two moving boundaries x = Si (t), i = 1, 2. Let S1 be moving towards +∞
and S2 moving towards −∞. Using a weak formulation of the above problem in Eqs (1.1.43),
(1.1.44), several interesting results on the behaviour of Si (t), i = 1, 2 have been obtained
in [10, 11]. The following proposition indicates that in some cases, the interface Si (t) starts
moving only after an elapse of time t∗ > 0.
Proposition 1.1.1. There exist numbers ti∗ ∈ [0, +∞) for i = 1, 2 such that Si (t) is strictly
monotone for t ∈ (ti∗ , +∞) and
If ti∗ > 0, then Si (t), i = 1, 2 remain stationary for ti∗ units of time [11].
In this case ti∗ is called a waiting time. It has been proved in [11] that the interface is
Hölder continuous under certain conditions and if the interface is in motion, then one expects
it to move with the velocity V of the gas, and
dSi m
= − lim Wx (x, t), i = 1, 2; W= ρ m−1 . (1.1.46)
dt x→Si (t)± m−1
The limits in Eq. (1.1.46) are taken as x approaches the boundary of the region from within the
region in which ρ(x, t) > 0. Eq. (1.1.46) can also be obtained from the mass balance condition
at the interface which states that the jump in the density at the interface multiplied by the
velocity of the interface is equal to the jump in the mass flux across the interface.
In a problem symmetrical with respect to x, it is sufficient to consider the region
0 ≤ x < ∞ with a single interface x = S1 (t) and a1 = a2 > 0. S1 (t) should satisfy
Eq. (1.1.46) for i = 1, and another condition to determine the unknown S1 (t) may be
prescribed as
S1 (t) a1
ρ(x, t)dx = ρ(x, 0)dx. (1.1.47)
0 0
Conditions of the type Eq. (1.1.47) are called nonlocal boundary conditions at the free
boundary. x = 0 is now a fixed boundary and the boundary condition on it is given by
∂ρ
= 0. (1.1.48)
∂x x=0
Note that we have two conditions prescribed at the free boundary, viz., Eqs (1.1.46), (1.1.47).
8 The Classical Stefan Problem
Problem 1.1.6 (Shock Propagation). The solution of Burger’s equation (1.1.49) with
boundary condition (1.1.50) has been discussed in [12]
1 ∂ 2
uy + uux = uy + (u ) = 0, (x, y) ∈ R2 , y ≥ 0. (1.1.49)
2 ∂x
On y = 0,
where
⎫
f (x) = 1, x ≤ 0, ⎪
⎬
= 1 − x, 0 ≤ x ≤ 1, (1.1.51)
⎪
⎭
= 0, x ≥ 1.
dx dy du
= 0, =1 and = 0. (1.1.52)
dt dt dt
The characteristic curves and u(x, y) are shown in Fig. 1.1.1. It is clear from the figure that the
characteristic curves intersect in the region x ≥ 1, y ≥ 1 and therefore u(x, t) is not defined
(is not single valued) in this region. u(x, y) defined by Eq. (1.1.54) ceases to be the classical
solution of Eq. (1.1.49) in the region x ≥ 1, y ≥ 1 and Eq. (1.1.49) is satisfied in this region
only in the weak sense. It may be noted that the solution u(x, y) could be discontinuous in
some portion of the region even if f (x) is a continuous and differentiable function throughout
the region. If Eq. (1.1.49) is written as
∂P ∂Q 1 2
+ = 0, P = u, Q= u , (1.1.55)
∂y ∂x 2
then it can be shown (cf. [12]) that the weak solution is discontinuous across a curve which is
called a shock and whose equation is given by
The notation [f ] denotes the difference between the limits of a function f as the shock or a
surface of discontinuity of f is approached from both sides. Conditions of the type Eq. (1.1.56)
are called Rankine–Hugoniot conditions. In the present problem [P] = 1 and [Q] = 1/2, and
therefore the shock is given by y = 2x + d. The constant d can be determined if we know
the point from which the shock is emanating, for example, in the present problem this point is
(1.1) (see Fig. 1.1.2). Shock is the oldest form of the free boundary and its origin lies in the
study of gas dynamics [13].
If Eq. (1.1.49) is multiplied by un , n ≥ 1, then P and Q in Eq. (1.1.55) will change.
This change will result in the nonuniqueness of the shock and infinitely many shocks can be
obtained. For nonlinear hyperbolic equations of the type Eq. (1.1.49), further information in the
form of physical arguments involving stability, entropy, dissipation or continuous dependence
on the initial data is needed to ensure uniqueness (cf. [14, 15]).
y y
u=0
y = 2x−1
u=1 (1,1)
1−x
u=
0 (0,1) x 1−y
1−x
u=
1−y 0 x
Fig. 1.1.1 Characteristic curves and u(x, y). Fig. 1.1.2 Geometry of the shock.
There are some important differences between the present problem and the free boundary
Problems 1.1.1–1.1.5. In the previous five problems, the free boundaries can be identified
without much difficulty and the boundary conditions on them can be imposed after some
thought. It is a different matter that the classical solution may or may not exist. In the solution
of Eqs (1.1.49)–(1.1.50), there is no indication of a free boundary. It is only through the
construction of the solution that we come to know about the magnitudes of jumps in P and
Q. Eq. (1.1.56) can be obtained only through the weak solution and not through the classical
approach, which was followed in other problems.
Problem 1.1.7 (Free Boundary Associated With a Frictional Oscillator Problem). This
interesting free boundary problem has been reported in [15]. As shown in Fig. 1.1.3, a block
of mass m rests on a conveyer belt moving with a constant velocity V. The forces acting on the
mass are: (1) spring force with a spring modulus Sm , (2) prescribed force F(t) assumed to be
sufficiently smooth and (3) Coulomb frictional force with coefficient of friction μ. The motion
of the block with mass m is governed by the equation
d2 x dx
+ μmg sgn − V + Sm x = F(t), (1.1.57)
dt2 dt
10 The Classical Stefan Problem
where
sgn(z) = 1 if z > 0,
= −1 if z < 0, (1.1.58)
Spring Mass
F(t)
x
V
Conveyer belt
Problem 1.1.8 (Impact of a Visco-Plastic Bar on a Rigid Wall). A bar of length b made up
of visco-plastic incompressible material moving with constant velocity −V0 hits a rigid wall
at time t = 0. We consider a one-dimensional problem in which x-axis is taken opposite to
the initial motion of the bar and the rigid wall is taken at x = 0. After the bar hits the wall,
compressional stresses develop in the bar giving rise to visco-plastic flow of the material in the
region. If V(x, t) is the velocity of the bar for t > 0, then the gradient of V(x, t) or the velocity
of deformation can be expressed as follows (cf. [16]):
τ +τ0
∂V μ , |τ | ≥ τ0 , τ0 > 0, τ ≤ 0,
= (1.1.59)
∂x 0, |τ | ≤ τ0 .
The Stefan Problem and Its Classical Formulation Chapter | 1 11
Here, τ represents compressional stress, which is negative as the x-axis is oriented opposite to
the direction of motion of the bar, τ0 is the stress at the limit point and μ is the coefficient of
viscosity of the material.
It will be assumed that any disturbance is propagated over the whole bar instantaneously.
The equation of motion in the visco-plastic region can be easily obtained by using Newton’s
second law of motion. We have
∂V ∂τ ∂ 2V
ρ = =μ 2, 0 ≤ x ≤ S0 (t), 0 < t < t∗ , S0 (0) = 0, (1.1.60)
∂t ∂x ∂x
where ρ is the density of the material and x = S0 (t) is the interface between elastic and visco-
plastic regions. In the elastic region
∂V
= 0, |τ | ≤ τ0 , S0 < x < b. (1.1.61)
∂x
In obtaining Eq. (1.1.62) it has been assumed that the elastic portion of the rod moves like a
rigid body.
Using again Newton’s second law of motion at x = S0 , we obtain
dX
−M = τ |S 0 + F0 , (1.1.63)
dt
where M is the mass of the elastic part and F0 is the area of cross-section of the rod assumed
to be uniform. Since M = F0 ρ(b − S0 (t)), Eq. (1.1.63) becomes
dX τ0
=− . (1.1.64)
dt ρ(b − S0 (t))
In obtaining Eq. (1.1.64) it has been assumed that stress is continuous at the interface x =
S0 (t). Using the condition that V(x, t) is also continuous at x = S0 (t), we get
Also
∂
V(S0 , t) = 0. (1.1.66)
∂x
The functions V(x, t), S0 (t) and X(t) are to be determined. An approximate solution of the
above problem has been discussed in [16].
The free boundary in this problem arises due to the change in the physical properties of
the system.
Problem 1.1.9 (A Problem With Discontinuous Moving Boundary). Discontinuity of the
free boundary in multidimensional problems can be easily demonstrated geometrically. For
example, a piece of ice floating in water breaks into two pieces after melting for some time.
A discontinuous moving boundary in a one-dimensional problem does not commonly occur.
In [17] an interesting problem of discontinuous moving boundary which is associated with the
diffusion of moisture in a porous capillary tube of length unity is discussed. At time t = 0, the
portion of the tube 0 ≤ x ≤ x0 , x0 < 1 is filled with moisture and the remaining portion is dry.
The temperature T(x, t) of the moisture is less than the boiling temperature T = 0, except at
x = x0 where T = 0. The temperature of the dry air in some neighbourhood of x = x0 on the
right is greater than zero and in the remaining portion, the temperature of the air is less than
zero. There is a continuous flow of moisture into the tube at x = 0. At x = 1, the dry air is
getting heated, causing evaporation to take place and making the moisture advance into the dry
air. Let T1 (x, t) and T2 (x, t) be the temperatures of the moisture and the dry air respectively,
W1 (x, t) the concentration of the moisture, and x = S(t), the equation of the free boundary
which is the interface between moisture and the dry air.
Under certain initial conditions it may happen that at some instant of time, say t = m,
m > 0, the temperature in some neighbourhood δm of the moving boundary becomes less than
or equal to zero. In this case the moisture will advance with a jump into the dry part, i.e.
The problem is concerned with finding T1 , T2 , W1 and S(t). The formulation of this problem
is as follows:
Differential equations
∂Ti ∂ 2 Ti
= a2i , (x, t) ∈ Di , i = 1, 2, (1.1.69)
∂t ∂x2
∂W1 ∂ 2 W1
=d , (x, t) ∈ D1 , (1.1.70)
∂t ∂x2
D1 = {(x, t) : 0 < x < S(t), 0 < t < m} ∪ {(x, t) : 0 < x < S(t), m < t < t∗ },
D2 = {(x, t) : S(t) < x < 1, 0 < t < m} ∪ {(x, t) : S(t) < x < 1, m < t < t∗ }.
Here, a2 is the thermal diffusivity and d is the mass diffusivity. Eq. (1.1.69) is the heat
conduction equation and Eq. (1.1.70), the mass diffusion equation (Fick’s law).
Initial conditions
Here, K is the thermal conductivity and α is the heat transfer coefficient. The convective
boundary condition in Eq. (1.1.76) arises because of the discontinuity of temperatures at
x = S(t), t = m. Heat balance at x = S(t) implies
dS ∂T ∂T
C2 ρ2 T2 (S(t) + 0, t) + K2 2 (S(t) + 0, t) − K1 1 (S(t) − 0, t)
dt ∂x ∂x
∂W
= −d q1 (S(t) − 0, t). (1.1.77)
∂x
Here, C is the specific heat, ρ the density and q1 the latent heat of evaporation. Definitions of
different parameters have been given later in Section 2.1.3. The derivation of Eq. (1.1.77) is
based on the law of conservation of energy (see Section 1.4.7).
A sufficiently small real number ε > 0 exists such that
For t = m,
T1 (x, m+) = T2 (x, m−), and W1 (x, m+) = 0; S(m−) ≤ x ≤ S(m+),
(1.1.79)
m± = m ± 0,
The existence and uniqueness of this problem under suitable assumptions have been discussed
in [17] and sufficient conditions for the existence of a discontinuous moving boundary are
given in [18]. Discontinuity in the temperature and in the free boundary is a typical feature of
this problem.
Problem 1.1.10 (Penetration of Solvents in Polymers). Consider a slab of a glassy
polymer, such as methyl methacrylate in contact with a solvent, n-alkyl alcohol [19]. If the
solvent concentration exceeds a threshold value, say, q ≥ 0, then the solvent moves into the
polymer, creating a swollen layer in which the solvent diffuses according to Fick’s law for
mass diffusion. If W(x, t) is the concentration of the solvent in the polymer and x = S(t) is the
free boundary representing the penetration depth of the solvent in the polymer, then W and S
satisfy the following equations:
14 The Classical Stefan Problem
Wt − Wxx = 0, (x, t) ∈ Dt∗ = {(x, t) : 0 < x < S(t), 0 < t < t∗ }, (1.1.81)
λ0 ∂p0 ∂ 2 p0 ∂p
u0 = − ; α02 = 0, b < x < ∞, t > 0, (1.1.85)
μ0 ∂x ∂x2 ∂t
λ ∂p ∂ 2 p1 ∂p
u1 = − 1 1 ; α12 = 1, S(t) < x < b, t > 0, (1.1.86)
μ1 ∂x ∂x2 ∂t
λ ∂p ∂ 2 p2 ∂p
u2 = − 2 2 ; α22 = 2, 0 < x < S(t), t > 0. (1.1.87)
μ2 ∂x ∂x2 ∂t
The subscripts 0, 1, 2 stand for the water-filled, intermediate and oil-rich zones, re-
spectively. ui and pi , i = 0, 1, 2, denote the filtration velocities and pressures in the
three different zones. The constants λi , αi and μi , i = 0, 1, 2, denote the coefficients of
permeability, piezoconductivity and viscosity of the ith zone. The equation of velocity in terms
of pressure gradient is the result of Darcy’s law [8] and the pressure obeys the equation of
piezoconductivity.
Boundary conditions at x = b and x = S(t) follow from the continuity of velocities and
pressures and are given by
The Stefan Problem and Its Classical Formulation Chapter | 1 15
u0 = u1 , p0 = p1 ; at x = b, (1.1.88)
u1 = u2 , p1 = p2 ; at x = S(t). (1.1.89)
At the free boundary, an additional condition is required which is obtained from the ‘mass
balance’ consideration
dS
m(1 − δ1 − δ2 ) = u1 . (1.1.90)
dt
Here, δ1 and δ2 are the contents of water and the residual oil saturation, respectively in
the transition zone, and m is the porosity of the stratum. To complete the formulation of
the problem, suitable initial and boundary conditions for pressure should be prescribed. For
example in [21], the initial pressure in the entire stratum is assumed to be constant. Similarly
it is assumed that pressure has a constant value at the boundary x = 0. The zero reference
point for calculation of pressure is so chosen that p2 (0, t) = 0. With the above assumptions the
initial and boundary conditions can be written as
An approximate solution under the quasi-steady approximation, valid at the initial stages has
been obtained in [21] with constant parameter values. The boundary conditions (1.1.89) do not
contain the velocity of the free boundary.
Problem 1.1.12 (Obstacle Problem for a String). This problem belongs to a class of prob-
lems which have a variational inequality formulation. Three different types of formulations
of this obstacle problem are possible and their equivalence is discussed in Chapter 7. A new
notion of codimensionality-two of the free boundary is associated with this problem and this
will be discussed briefly in Section 1.2. Consider a weightless elastic string which is held
tight between two fixed points P1 = (0, 0) and P2 = (b, 0), b > 0 in R2 -plane. This string
is displaced upwards by a rigid body called an obstacle (cf. [15, 22]). A view of the cross-
section of this system of string and obstacle from above in R2 -plane is shown in Fig. 1.1.4.
Let y = ψ(x) be the equation of the cross-section of the obstacle in R2 , i.e. the equation of
the curve Q1 ABQ2 . ψ(x) is assumed to be sufficiently smooth. We shall relax this smoothness
condition later (see Section 7.2.5). Let y = u(x) be the equation of the string in the equilibrium
position or the equation of the curve P1 ABP2 . The problem is to find the function u(x) and
the arc AB of the string which is in contact with the obstacle and on which u(x) = ψ(x). The
points A and B are free boundaries and in the present case we can call them ‘free points’. Once
A and B are determined, the arc AB is also determined because ψ(x) is known.
The second equation in Eq. (1.1.93) implies that the string does not penetrate the obstacle. The
concavity of the string as viewed from the x-axis implies Eq. (1.1.94) and the tightness of the
elastic string implies that the portions P1 A and BP2 are straight lines and so u (x) = 0. The
two boundary conditions at the free boundary can be obtained from the continuity of u and
du/dx, i.e.
du du
[u]A = [u]B = = = 0, (1.1.96)
dx A dx B
where [f ] stands for the jump in the quantity under consideration at the given point. Eq. (1.1.95)
is equivalent to the following condition
In view of Eqs (1.1.93)–(1.1.95), Eq. (1.1.97) is valid. When u(x) = ψ(x), Eq. (1.1.97)
is satisfied and when u(x) = ψ(x) then the second equation in Eq. (1.1.93) implies
u(x) > ψ(x) and from Eq. (1.1.95), u (x) = 0 and therefore Eq. (1.1.97) is satisfied. If
Eq. (1.1.97) holds and if u(x) > ψ(x), then u (x) = 0, which is Eq. (1.1.95). The formulation
(1.1.92)–(1.1.95) is equivalent to the formulation given by Eqs (1.1.92)–(1.1.94), (1.1.97).
Let us call the formulation (1.1.92)–(1.1.95) as Problem (R1 ) and the formulation equivalent
to Problem (R1 ) as Problem (R2 ). Consider the following minimization problem which is
concerned with the minimization of the energy of the above string:
P2 2
du
min dx, ∀u(x) ∈ D, (1.1.98)
u≥ψ P1 dx
where D = {u : u(0) = u(b) = 0, u and ∂u/∂x are continuous, and u satisfies relations
(1.1.94) and (1.1.95)}.
We shall call this formulation in Eq. (1.1.98) as Problem (R3 ). This is a fixed domain
formulation as it does not consider separate formulations on different portions of P1 ABP2 . In
view of the equivalence of Problems (R2 ) and (R3 ) which will be established in Chapter 7, the
formulation (R2 ) is also called ‘variational formulation’ of the obstacle problem. A variational
inequality formulation of this problem has been given in Chapter 7.
y
y = Y (x)
A B
String
P1A
P2B
P2 = (b,0)
x
Obstacle
P1 = (0,0)
Q1 Q2
CHAPTER IV
AN IDYLLIC LIFE
When Sandie awoke again, he felt so much fresher, lightsomer, and better,
and was admitted by the doctor to be so far recovered that he was permitted
to sit up a little and engage in conversation with his mother and gentle little
nurse, Maggie May.
The latter interested Sandie very much indeed. He had never before seen
a child-girl half so lovely. To him she was idyllic, a poem, a dream-child. It
seemed to this romantic ploughboy-student as if Maggie May—what a
sweet name, too!—had flown straight out from the pages of Anacreon.
Of course there may have been a good deal of super-sentimentality about
all this, for the mind is always more sensitive when the body is feeble and
weak; and weak Sandie still was, and would be for many a day. However, it
may be confessed, before we go any farther, that Maggie May was an
innocent, artless, and a very beautiful child.
I have myself an opinion that no girl can be really beautiful who is not
truly good, whose heart is not imbued with religion and in touch with
nature. If the soul, in all truthfulness, does not shine through the eyes, be
they brown or be they blue, then, ah! me, beauty is far, far away. And yet
many girls now-a-days think that the more closely they approach in figure,
face, and complexion to the waxen dummies we see in the windows of
hairdressers the prettier they must be. A greater mistake could not be made.
Let me say earnestly to every girl who may read these lines, “Cultivate
mind and soul if you wish to become beautiful.”
This is a digression, and I apologise for it, and proceed with my true
story.
A day or two afterwards, Sandie’s sister came over to the manse, and the
mother went home.
Maggie May and she soon became fast friends, and together it was
evident they would soon nurse Sandie back to life.
Maggie May possessed a zither, on which, for so young a girl, she
played charmingly, singing thereto old Scotch songs, such as “The Flowers
of the Forest,” “The Parting,” “Wae’s me for Prince Charlie,” and other
Jacobite lilts, that caused the tears to come welling up into Sandie’s eyes till
he could see nothing for the mist they produced; for Sandie was still very
weak and hysterical.
The minister came daily, twice a day, to see the patient. One day he
brought Sandie’s Horace.
“Do you mean to tell me, Sandie,” said the minister, “that you read
Latin?”
“Oh, yes, just a little. And a little Greek,” he added.
Mackenzie patted his thin white hand, and looked wonderingly down
into his pinched and worn face.
From that moment Sandie knew he had found a friend.
Then he told him all—all his ambitions, all his struggles, and all his
doubts and fears.
Mackenzie was silent for a time after he had ceased speaking. Then he
took Sandie’s hand in his. “Listen!” he said. “I was a bursar at my
University, or I would not be where I am now, for my people were only
fisher-folks at Peterhead. I was a bursar, and I have ever since kept up my
classics. Now, I can put you in the way of working up for the Grand
Competition at the end of October, if you care to come over here about
twice or thrice a week.”
Once more came that wildering mist of tears to Sandie’s weak eyes.
“The Lord be praised and you be thanked,” he said, pressing Mackenzie’s
hand. “He has raised me up a friend, and I am more happy now and hopeful
than I have ever been in life.”
For another whole week Sandie was still so weak as to be unable to
leave his room; then he was able to totter out into the minister’s garden, and
seat himself on the summer-seat, in the warm spring sunshine, in the
healthful bracing breath of the sweetest month in all the year.
Maggie May went with him, and sat near him, and read to him little
stories, in which he pretended to take great interest, though it really was the
story-teller, not the story, he was studying all the time. Soon after his first
out-going, young blood began to assert itself, and he somehow felt ashamed
of being ill or a patient. He was getting rapidly stronger, at all events, and
one morning announced his intention of going home. The minister knew it
would be useless to argue with him. Genius is wilful, and there was every
probability even now that Sandie would eventually prove that he possessed
genius. “What is genius after all,” said somebody, or words to this effect,
“but the capability of plodding and steady work?” I am certainly not
prepared to agree with this. Genius depends greatly on brain power and
brain formation. I never would expect much except a grunt from a sow,
however much she applied herself to study.
Sandie went home. The spring and merry May were now almost gone.
The joy of June would soon be here. The men, and even Jeannie, the simple
servant lassie, were busily engaged thinning the young turnips. As Sandie
drove slowly down the loanings in the gig, he could hear their merry voices
as they talked and laughed, with now and then Jeannie’s gentle voice raised
in song, to which Jamie appended a deep broad bass. The horses were still
in the fields as he had seen them last—Glancer nibbling the shoulder of
Tippet, Tippet nibbling that of Glancer, the best proof one horse can give
his fellow that he loves and respects him.
The banks by the dike and ditch-sides were now all ablaze with the most
charming wild-flowers. I might be accused of making copy were I to
mention the half of them; but on the water itself floated the spotlessly white
water-anemone and the wild forget-me-not. On the banks near by nodded
the crimson ragged-robin and blood-red selené. They seemed to be looking
at and admiring their own sweet faces reflected from the pools beneath. But
the banks were also patched with sky-blue speedwells, starred over with
great, solemn-looking, oxeye daisies, and backed by a profusion of the tall
and lordly purple orchis.
Sandie took all this in at a glance. His own humble home was the chief
part of the picture before him; the banks of wild-flowers, and the clear
flowing wee burns or streamlets, were but settings.
His doctor had warned him that he must not use his study for some days
to come. Sandie had promised, and he determined to obey. Well, he could
not work just yet, so he determined to fall back upon Robbie Burns and
Anacreon. With a volume of each in his pocket, he went to the fields every
day, and just dawdled along behind the workers, the rooks in turn following
up at a respectful distance behind all. Sandie read to the workers, and read
so pleasantly, that one moment he would have all hands laughing enough to
scare the very rooks, and next the men-folks looking solemn and sad, and
the salt, salt tear in Jeannie’s eyes. Dear me! what a power there is in poetry
and song when it is well and feelingly read! Somehow I cannot help
thinking that, to read poetry well, the reader himself must be possessed of a
portion of the divine afflatus.
“Well, mother,” said Sandie one evening, just after June had come in,
“I’ve made up my mind to go in for the bursary competition in the end of
October. I can but fail.”
“You winna fail, laddie. I’ll pray.”
“Ah! mother, prayer is only one thing. I’m going to work.”
“You winna kill yoursel’?”
“No fears, mother. Honest work never killed anybody, though the hoofs
of a daft Shetland pony skilfully applied might. No; I’m going to work,
mother mine, and go over twice a week to see Minister Mackenzie. It really
is good of him to promise to put me on the straight road, isn’t it?”
“It is, laddie. It was mebbe all for the best that the pony hurt you.”
“I think it was.”
“God moves in a mysterious way, Sandie.”
“He does, mother; but now there is something else worrying me. Should
I succeed in getting a bursary, that, with the addition of a little pupil-
teaching, will be enough to support me, won’t father miss my work very
much all winter?”
“We maun do the best we can, laddie; that maunna stand in the way o’
your advancement. Na, na, Sandie; banish a’ sich thochts frae your heid.”
“Weel then, mother, I’ll make my first run over to the minister’s to-
morrow, and to save time I’ll ride on Lord Raglan. He’ll be turned into one
of Mackenzie’s fields till I’m ready to come back.”
. . . . . .
That was one of the most pleasant day’s outings that ever Sandie had
had, and there were many such to follow during the long sweet summer
days.
Mackenzie was simply astonished at the amount of the lad’s erudition.
He, however, managed to put him right in many little things; that is, there
were subjects that Sandie had been studying, and studying hard too, which
would not be required of him while competing for a bursary. It would be
obviously worse than useless to continue with these. So the minister was of
real service to our ploughboy-student.
But Mackenzie was wise in his day and generation. No one knew better
than he that a brain kept constantly on the rack soon becomes a weakened
brain, and that poverty of blood and body follows. So on the days when
Sandie came over to the manse, the kindly minister just granted him three
hours of tuition in the forenoon; then came luncheon, and after that he was
sent off to fish. On these little piscatorial forays, Sandie’s constant
companion was little Maggie May. None knew better than she where the
best
“HE WANTED TO WATCH MAGGIE MAY”—Page 37.
and biggest mountain trout lay, or where to use fly and where to fish with
bait; and her knowledge she invariably communicated to her big
companion. And he—well, he never had been very much of a fisherman,
but now it seemed to him that he was less artistic than ever. If the truth must
be told, he could not do so much as he could have wished, because he
wanted to watch Maggie May. There was something in every look and
movement of this beautiful child, and in her innocent prattle as well, that
drew Sandie irresistibly towards her. To his way of thinking she was idyllic.
Was he falling in love with the bonnie bairn? Oh, I do not wish for a
single moment to suggest anything of the sort; only be it remembered that
Sandie really was a poet at heart, and that poets love all things lovely that
they see around them.
Towards six o’clock sport always ended, and with their bags on their
backs, and fishing-rods over their shoulders, they went together slowly back
to the manse.
Dinner followed. Mackenzie would always insist on his pupil staying to
dinner. Then, in the calm summer’s gloaming, Sandie would bid his friends
adieu, mount Lord Raglan, and ride slowly home. Mrs. M‘Crae and his
father invariably sat up for him, and he had always much that was hopeful
to tell them. But he must even yet spend a few hours in his study; for,
pleasant though they were, Sandie could not help looking upon those
fishing excursions as so much time wasted or thrown away. Therefore he
resorted to his rustic study in the corn loft, and there he would sometimes
sit till grey daylight in the morning. This at the summer’s height is not
necessarily very late, for, far away north in Aberdeenshire, about mid-
summer, there is really very little darkness.
But never, I ween, did sleeper sleep more sweetly than did Sandie when
his head was at last on the pillow. Slumber stole over his senses—
immediate, instantaneous—and he never awoke until Tyro the collie put his
paws on the bed and licked his ear; and thus for the present was his life
almost an idyllic one. Alas! this is a kind of life that does not last long with
any one in this weary world.
CHAPTER V
SORROW NEVER COMES SINGLY—CHRIST-LIKE
CHRISTIANITY
I don’t think there is a more truthful aphorism in our language than that
which tells us that sorrows seldom come singly.
Fortune or fate had dealt so very hardly with honest Farmer Kilbuie last
season, that he might reasonably have expected now some surcease of
sorrow—a respite, if not indeed a flow of good luck. Alas! it was otherwise.
The turnips had been thinned and earthed up—they were already
beginning to cover the drills—and the haymaking season was in full blow.
It was hot sunshine now every day, with now and then a gentle breeze
blowing from westward or south, a breeze that blew through the tossed and
tumbled hay and made and “won” it.
There was still a good deal to cut down, however, and Sandie himself
was walking behind the reaping-machine with the great horse Glancer
dragging. This machine not only cut the hay, but tossed it into wreaths.
Sandie didn’t look particularly like a student or genius at present. He
wore little save a blue checked shirt, his trousers, and a wide-brimmed
straw hat, inside which was a cabbage leaf as a security against sunstroke.
The mowing went merrily on.
In another part of the field the servants, with Mr. M‘Crae himself, were
busily and cheerfully engaged among the hay that had been cut down
yesterday, and which was already dry enough to put into “cocks” or “coles.”
Sandie was just about half-way down a ridge, when he pulled up to wipe
his wet perspiring brow. Just at that moment Glancer threw up his head and
emitted a kind of pained and stifled cry. He reeled for a moment, then fell
heavily on his side. Coup de soleil, or sunstroke, without a shadow of
doubt.
Mr. M‘Crae and the servants saw the poor horse fall, and hurried at once
to Sandie’s assistance. At first an attempt was made to raise the animal, but
this was found impossible; the neck drooped, the legs were paralysed.
M‘Crae had always been his own veterinary surgeon, and perhaps knew
quite as much about the ailments of cattle and horses as did the drunken
little smith and farrier who lived in the neighbouring village. So Glancer’s
harness was unloosened, a bundle of soft dry hay was placed under his
head, and a canvas shelter was erected to save him from the burning rays of
the sun. His poor head, too, was kept constantly wet with the coldest of
water, and now and then his tongue was pulled to one side, and a cooling
draught administered.
Sandie and Jamie never left him all that day; Jeannie brought their
dinner out to the field, and their supper also, and they ate it beside the dying
Glancer.
Poor Tyro, the collie, seemed to know he was in the presence of death.
He sat or lay, though not asleep, near to the horse till the end, often heaving
deep sighs, for the farm nags were all special favourites of his.
Tyro really was a faithful and kind-hearted dog. I need not tell the reader
he was wise, because he was a Scottish collie, and collies are the kings of
the race canine. Yes, he was loving and gentle, and he was an excellent
guard by night. Once upon a time he surprised a hawker-tramp robbing the
fowl-house. Tyro did not fly at the man and bite him, as a less sensible dog
would have done. No, he simply placed that fowl-house, with the itinerant
hardware merchant inside, in a state of siege.
“If you dare to come out,” Tyro told him, “I will cut your throat, as
certain as sunrise.”
So the unhappy man preferred capture to a cut throat; and when M‘Crae
came round in the grey dawn, he found the tramp, and in due course he was
landed in prison.
But in the interests of truth, I must state here that Tyro had one fault, and
a very sad one it was. In company with another dog, a smooth-coated cross
’twixt a greyhound and collie, he used in the season to go hunting the
turnip-fields for hares or rabbits. They worked very systematically, Spot
going into the field to start the game and chase it towards the gate, where
Tyro lay in wait to seize and kill it. In this way they sometimes laid dead as
many as six or eight hares a night, bringing home one each in the grey of
the morn, and hiding the others to be recovered by degrees.
Tyro had even been accused of sheep-killing, but the crime was brought
home to another dog, and Tyro left without a stain on his character.
Just as the sun had dipped behind the wooded hills of the west, and
gloaming shadows began to fill up the hollows, it was evident that great
Glancer’s minutes were numbered. The fast glazing eye and the stertorous
breathing told the watchers that. Soon after, he had a few fits of shivering,
one last long sigh, and then he lay still—all was over.
Jamie Duncan had kept up till now, but when he heard that sigh, and
knew the horse was dead, he lost all control over himself, and threw himself
on the body in a paroxysm of grief and tears.
You must remember he was an illiterate ploughman, reader.
“O Glancer, Glancer!” he cried; “oh! my poor dead friend Glancer, will I
never mair clean your harness, or lead you to the fields in the mornin’? O
Glancer, my heart is br’akin’! my heart is br’akin’!”
And so he kept on for a time, until Sandie insisted on leading him
homewards.
But Jamie wasn’t well for days.
The next death at Kilbuie occurred about two weeks after this, and
affected Mrs. M‘Crae and her two children more than any one else. It was
that of Crummie, a cow nearly fifteen years old, but yet in calf. She took
what is called the “quarter-ill,” or mortification of one joint or limb, and
quickly succumbed. There was a halo of romance about this wise old cow.
Like the bovine in the old Scotch song called “Tak’ your auld cloak about
you”—
Ah! that was just where the sorrow came in. Long, long ago, when
Sandie and Elsie were but toddling thingies, in the bright and early days of
her husband’s love, when all was hope and happiness about the smiling
farm, and sorrow seemed very far away indeed, that old-fashioned cow had
given the milk for the bairnies’ porridge, and the cream for butter. During
all these long years she had kept the same stall in the byre, and woe be to
any other cow beast that thoughtlessly dared to enter it. The retribution was
sharp and swift.
Hardly ever a day passed either that, before going to her stall, after
having been out for water or away in the green fields, Crummie did not
come to the back door and knock with her head, and Mrs. M‘Crae, or
Jeannie latterly, would present her with a nice piece of oat-cake, after which
she would gracefully retire, that is as gracefully as a cow can, walking
backwards a considerable way, as if she had been in the presence of royalty.
But now Crummie was “nae mair,” as Jeannie phrased it, and the bairns
and the mother were inconsolable.
In a week more the calf would have been born. As it was, its skin was
utilised. There is a curious but rather beautiful superstition away in northern
Aberdeenshire, namely, that the very large family or hall Bible should be
covered with the skin of a calf that has never been born. So poor Crummie’s
calf’s skin was used by M‘Crae to cover his great Brown’s Bible.
. . . . . .
Now I must tell you that Kilbuie was very much respected and beloved
by the neighbouring farmers. For Kilbuie was a farmer, and not an upstart.
He had been among them all his life. His father, too, had farmed Kilbuie
before him. Had M‘Crae been a shopkeeper or sailor turned farmer, they
would have left him severely alone. They were clannish.
Well, one evening there was a secret meeting of these farmer folks in the
little village school-house. It was a secret meeting, but they weren’t plotting
to blow up the manse with dynamite, or set the old town-hall in a blaze. No,
and the result of the secret meeting one day about a week after walked
down the long loaning towards Kilbuie, in the shape of a fine sturdy young
cart-horse, as like Glancer as possibly could be. He was, as may be guessed,
a gift to the unfortunate M‘Crae from his kindly neighbours. To refuse
would have been to offend. So what could he do but accept, to thank and
bless them? The neighbours’ kindness did not end here. They had heard that
Sandie M‘Crae meant to compete for a bursary, and, after taking his Master
or Bachelor of Arts degree, study for the ministry. Well, it occurred to them
that, one way or another, Kilbuie would be rather short handed for the
ensuing harvest, that is, if Sandie was going to get anything like fair play,
and be allowed to make preparations for the competition; so they
determined to give Kilbuie a love-darg, not only for the harvest, but with
the subsequent ploughing.
In case there may be some readers of mine in the far south who do not
know what a love-darg means, I must explain. I have said already that the
farmers of the North are clannish. Well, it often occurs that when, through
misfortune, one of their number falls behind-hand, say in the ploughing, the
neighbours all assemble in force with horses and ploughs, and in one day
turn over every yard of his stubble or leas; or in the same way they may
sow his oats in spring, or reap them for him in harvest-time.
Surely this is genuine and Christlike Christianity!
They did not, however, communicate their intention to the farmer
himself, but to Sandie they did. Sandie’s eyes sparkled with joy.
“Hurrah!” he cried, “the bursary is as good as won. How can I thank
you, gentlemen?”
“By no thankin’ us at a’,” returned Farmer Mon’ Blairie, the spokesman.
“Man!” he added, “we’re a’ as prood o’ ye, lad, as prood can be. We’d
like to hae a minister reared frae among oursels, and we’ll hae you.”
“I hope so.”
“Weel, keep up a good heart. Ye can study a’ the hairst.”
“I’m going to do something else besides.”
“Weel?”
“Ye see, if I can manage to get just one month at the Grammar School of
Aberdeen before the competition, it will ensure my success.”
“To be sure; weel?”
“Weel, by the merest chance yesterday I met Lord Hamilton at the
minister’s manse. He was having lunch there. He was bemoaning the fact
that when the grouse-shooting began on the Twelfth, he should not have a
single keeper who thoroughly knew the hills. Then a happy thought
occurred to me, and something made me speak.
“ ‘My lord,’ I said, smiling, ‘there isn’t a corrie nor a knowe, a height
nor a howe, all over these hills that I haven’t known since my childhood;
will you accept my services as your head-keeper? I’ll serve you well and
faithfully till past the middle of September.’
“ ‘But you,’ cried his lordship, laughing, ‘the minister’s friend and a
farmer’s son! I should never think of offering you a post so menial. Oh! no,
boy; you must be joking.’
“ ‘But I’m not joking,’ I insisted.
“Then I told him all the truth, and all my ambition to win a bursary and
to study for the ministry, and to do all and everything by my own exertions
entirely.
“He smiled once more; then he stretched out his soft white hand and
grasped mine.
“ ‘Sandie M‘Crae,’ he said, ‘I admire your pluck; you’re a Scotsman
every inch. Yes, I accept your services. Be at the shooting-box the day
before the Twelfth.’ ”
. . . . . .
The Twelfth of August—that glorious day on Scottish hills—came round
at last, and Sandie found himself starting off to the heather with Lord
Hamilton and party long before sunrise. There was to be no battue shooting,
none of that unfair driving so common in Yorkshire: each man walked
behind his well-trained setter and retriever. This was real sport, and gave
the birds a chance, as well as showing what kind of a shot each man was.
Sandie attended personally on Lord Hamilton, and gave such entire
satisfaction that his lordship was loud in his praises at eventide, when he
found his bag so large that two ordinary keepers were needed to carry it.
There was a great dinner-party that day in the shooting-box, and wine
and wit sparkled bright and merrily; but Sandie, as soon as he had dined
sumptuously in the kitchen with the other keepers, begged leave to retire,
and sought the solitude of his little bedroom, where his books were, there to
study as usual till far into the night.
He was up and ready for Lord Hamilton, however, some time before that
gentleman appeared, and another excellent day on the hill succeeded.
Well, why need I say more about it? Each day was like another, and so
the time flew on, only Sandie grew every day more brown and hard, till at
the end of the six weeks he left Lord Hamilton’s service as happy as a king,
with his lordship’s words of praise ringing in his head, and quite enough
money jingling in his pocket to maintain him for a whole month and a week
at the Grammar School.
CHAPTER VI