3

ELECTROMAGNETISM AND FIELDS

Which is more real, fields or matter? This question often arises, but
it is clear that the two views complement and reinforce each other.
The discussion so far has emphasized the material point of view. In
this chapter the field description is used, and partial differential
equations are obtained governing the distribution of magnetic field
in space. The field description then becomes the basis for deriving
an expression for the state of stress in vacuum, the remarkable
Maxwell stress tensor. Subsequently, with this understanding it will
be possible to advance in Chapter 4 to the analysis of stress in
magnetized media. For that purpose the complete Maxwell’s
equations of the electromagnetic field will be needed, as well as
certain expressions for energy density that are derivable from them.
This treatment begins with another look at Coulomb’s law.

3.1 Magnetostatic field equations

Coulomb’s law of magnetism was introduced as equation (1.1).
The H field generated at position r by a point pole p located at the
origin of coordinates can be written

Figure 3.1 represents the field piercing a surface s of arbitrary

shape surrounding the pole.
 3.1 An imaginary closed surface S that encloses a point pole at the origin. This diagram
is used with Coulomb’s law to derive the equation ∇.B = 0.

Because H can be regarded as the density of field lines, the flux of

field lines through a surface patch of area dS is given by

where dΩ = r·dS/r3 is the solid angle in steradians subtended by dS .

Note that only the solid angle appears in the rightmost term, and not
the radius, so the enclosing surface may be of any shape. The flux of
field out of a closed surface is then

because the solid angle subtended by a whole sphere is 4π

steradians. For N poles inside S, where the ith pole contributes an
amount Hi, to the total field, a sum of such equations gives

and bringing the summation inside of the integral on the left side
results in

where it is recognized that the resultant magnetic field H is the

vector sum of the individual contributions, and pi can take on
positive or negative values. This result can be immediately
generalized to the case of a continuous distribution of poles
characterized by a pole density ρv. The number of poles in the
element dV is ρvdV, so

where V is the volume enclosed by the surface S.

Applying the divergence theorem to the left side of the equation
and noting the arbitrariness of the volume V yields the differential
equation

which applies in all regions of space whether magnetized matter is

present or not. In the addendum to Chapter 1 it was shown that
volumetric pole density and if this is compatible with
the expression given here for ∇·H, it must be true that ∇·H = −∇·M.
Previously, the vector B was defined such that

and so it follows that B satisfies the relationship

Equation (3.3) is one of Maxwell’s equations, and the vector B,

previously introduced in Chapter 1, is known as the magnetic
induction.
From the form of (3.3) it follows that B is analogous to the velocity
vector of an incompressible fluid. Thus, the B field can be pictured
as the flow of an incompressible fluid. The amount of the fluid
entering an arbitrary volume equals the amount flowing out - none
accumulates. Also, the lines of B cannot terminate but must form
closed loops or extend indefinitely far.
Additional information can be obtained from Coulomb’s law. From
(3.1) it may be inferred that the field at a position r due to a
collection of poles Pi located at positions ri can be written

Next it is desired to find the curl of H, which involves evaluating a

sum of terms of the type
The right side of this equation results from applying the vector
theorem for the curl of the product of a vector and a scalar function
(see Appendix 1). By direct calculation,

Because the cross product of parallel vectors is zero, it follows that

This result represents the magnetostatic form of Ampère’s law,

which is applicable when there is no current flow.
Thus, it has been shown that the magnetostatic relations ∇·B = 0
and ∇ × H = 0 are mathematical consequences of Coulomb’s law.

Scalar potential
Because the curl of H is zero, a scalar magnetic potential ψ
can be defined such that

and (3.4) is satisfied identically. With B proportional to H, or for

uniform M, the divergence of H is zero, which follows from (3.2) and
(3.3). Thus, taking the divergence of both sides of (3.5) shows that
the scalar magnetic potential obeys Laplace’s equation,

for which many solutions are known from problems in electrostatics,

thermal conduction, molecular diffusion, and elasticity. An
illustrative solution of equation (3.6) will be developed in Section
3.2, after the general boundary conditions that the fields must
satisfy have been developed.

3.2 Magnetic-field boundary conditions

The divergence theorem gives an integral representation that
is equivalent to the relationship ∇·B = 0.
Similarly, Stokes’s theorem applied to ∇× H = 0 results in Ampère’s
circuital law:

which is valid when there is no current flow.

At interfacial boundaries separating materials of different
properties, the magnetic fields on either side of the boundary obey
conditions that are determined by (3.7a), and (3.7b), applied to a
differential volume straddling the interface. Figure 3.2 shows at the
left a small volume (the Gaussian pillbox) whose upper and lower
surfaces are parallel and are located on either side of the interface.
The short cylindrical side, being of zero length, offers no
contribution in the first integral relation, (3.7a), which thus reduces
to

where the subscript n denotes the normal component. This result

states that the component of B normal to the interface is continuous.
The result may also be expressed

Ampère’s circuital law may be applied to the contour of differential

size enclosing the interface, as shown on the right in Figure 3.2.
Because the sides labeled BC and AD approach zero length, they
offer no contribution to the integral. The remaining two sides yield

Thus, the tangential component of magnetic field is continuous. This

may be written
 3.2 The normal component of B and the tangential component of H are continuous
across the interface between dissimilar regions.

The use of these relationships will now be illustrated.

Example - magnetic slab within a uniform, applied magnetic

field: A slab of infinite extent in the x and y directions is placed
within a uniform magnetic field H0 having orientation normal to the
slab boundaries, as shown in Figure 3.3. Find B, H, and M within the
slab when it is
(a) permanently magnetized with magnetization M0 = M0k, and
(b) a permeable material with constant permeability μ ≡ B/H.
Solution: For both situations, (3.8) requires that the normal
component of the magnetic induction B = μ0H0 be continuous across
the boundaries.
(a) Thus, for the permanently magnetized slab, the continuity of B
across the boundaries requires that

B0 = B

Therefore,
 3.3 Uniform magnetic field applied to slabs of hard and soft magnetic materials.

By definition, M0 is constant. Note that when there is no externally

applied field (H0 = 0), the resultant field within the slab is oppositely
directed to the magnetization and equal to it; i.e., H = −M0. The
field within the slab disappears when H0 = M0.
(b) For a linearly permeable medium, (3.8) requires that

μH =μ0(H + M) = μ0H0

Eliminating H gives the magnetization induced in the slab:

M = H0(l − μ0/μ)

This expression shows that, no matter how large the value of the
permeability μ, the magnetization M cannot exceed the value of the
applied field H0. The geometry prevents the slab from becoming
highly magnetized.
Example - sphere in uniform magnetic field: In spherical
coordinates with axisymmetric symmetry, Laplace’s equation for the
potential has the form

where r is the radial distance from the origin and θ is the polar
angle. A trial solution in the product form

leads to solutions in terms of Legendre polynomials, only the first of

which suits the conditions of the problem. This yields as the form of
the solution,
where R is the sphere radius. The associated magnetic field is then

The magnetic field far from the sphere must approach the uniform
applied field H0k, where k is the unit vector in the z direction. In
spherical coordinates k = ircos θ − iθ sin θ, so for r →∞ it may be
seen from the solution for H that

By the same token, the form of the solution for r < R shows that H
=Ak, and hence the magnetic field within the sphere is uniform and
z directed. The solution outside the sphere is the imposed field plus
a contribution as if there were a magnetic dipole at the center of the
sphere with moment 4πD.
Because the tangential component of H and the normal component
of B are continuous across the spherical interface, two additional
relationships are obtained for the unknown constants. These yield
the solutions

where μ1 is the permeability of the medium surrounding the sphere,

and μ2 is the permeability of the sphere. The magnetic field lines are
plotted in Figure 3.4 for the case of a sphere permeability
approaching infinity. In this limit, H within the sphere is zero, and
the field lines incident on the sphere are purely radial.
 3.4 Magnetic field lines about an infinitely permeable sphere subjected to a uniform
impressed magnetic field.

3.3 Maxwell stress tensor

The physical and mathematical meaning of the stress tensor
was developed in Section 1.7. As will be shown, the notion provides
a powerful alternative description of magnetic force transmission to
the force-at-a-distance concept inherent in Coulomb’s law. A
mathematical expression for the magnetic field stress tensor is
developed next. It should be kept in mind that no material medium is
required to transmit magnetic stress.
To begin it is helpful to visualize a small “test” cloud of magnetic
dipoles surrounding a point in space. From Brown’s theorem,
equation (1.63), the cloud of dipoles is equivalent to a pole density
distribution ρv. Because H is the force per unit pole, the apparent
local force density F is

What will be done next is to manipulate (3.10) so that the right side
can be written as the divergence of a tensor, the Maxwell stress
tensor. By(1.63), or πv = − μ0∇·M, and by equations (3.2) and (3.3),
or −∇·M = ∇·H, (3.10) becomes

With the aid of the tensor identity ∇·(HH) = H(∇·H) + H·∇H, (3.11)
can be written
However, H·∇H = ∇(1/22H) − H × (∇ × H) and, with the relationship
∇ × H = 0 [equation (3.4)], (3.12) can be rearranged to become

It is shown presently that for any scalar function s, and

thus (3.13) can be written as the divergence of a tensor in the form

where the quantity in the parentheses is the Maxwell stress tensor

T:

or, in indicial notation,

From the mathematical properties of a stress tensor, the traction

force on an arbitrary patch of surface dS having the outward-facing
normal n is n·TdS, and so the total force acting on a body occupying
volume V enclosed by surface S may be written

The importance of this result is that the magnetic force acting on the
contents of a “control” volume can be evaluated using only
knowledge of the field existing over the surface of the volume. The
shape and position of the volume may be chosen for convenience in
evaluating the terms of the expression. An illustration of the utility
of the concept is given in Appendix 2 as the analysis of a sheet jet.
 It should be noted that although the Maxwell stress tensor was
constructed on the assumption of a magnetic force exerted on a
distribution of dipoles, the result is applicable in free space, i.e.,
where all matter is absent. That is, the moment of the dipole cloud
can be assumed to vanish. Thus, in magnetostatics the integration of
the Maxwell stress vector n · T over a closed surface of empty space
must yield zero force.

Portrait of the Maxwell stress tensor

Consider an element dSof a surface S, shown in Figure 3.5. It
is desired to compute the traction acting at point P where the unit
normal n makes an angle θ with the x axis:

3.5 Development of equation (3.21).

For argument’s sake assume that the coordinate system is chosen

such that the magnetic field evaluated at P is oriented in the x
direction; i.e., H = Hi Let the angle that the traction vector tn makes
with the x axis be denoted ϕ Then tn ≡ n·T = T·n (symmetric T) is
given by

Using (3.18) and H = Hi, one finds that

Thus

Since ty/tx is also equal to tan ϕ, it may be concluded that

This proves a very general result: In words, the angle between the
surface normal n and the traction vector tn is always bisected by the
magnetic field H. From the foregoing analysis it is seen that the
traction vector corresponds to collinear tension when H is parallel to
n and to transverse pressure when H is perpendicular to n. These
facts are illustrated in Figure 3.6, where . Viewed in another
way, the Maxwell vacuum stress tensor gives rise to tension in the
direction of the field lines and to pressure in the perpendicular
direction; that is, the lines of force behave like rubber bands. A
careful study of Figure 3.7 should clarify some of the points made in
this section.

3.6 Geometrical analysis of the stress vector tn: (a) Collinear tension and (b) transverse
pressure.
 3.7 Portrait of the Maxwell vacuum stress tensor in a magnetic field H. As the surface
normal vector n rotates clockwise, the stress vector tn revolves counterclockwise.

3.4 Maxwell’s equations

The historical development of electromagnetic theory followed
two separate paths until the nineteenth century. One of these was
the study of electric charges and their fields, and the other
concerned electric currents and the magnetic fields they produce.
This was the state of affairs until Faraday showed that a time-
varying magnetic field can generate an electric field and Maxwell,
through introduction of the displacement current, showed that a
time-varying electric field produces a magnetic field. The
mathematical relationships governing electromagnetic phenomena
are the celebrated Maxwell’s equations. We cannot hope to trace the
development of the full equations in a short compass, but we shall
state these laws and discuss aspects of them. The predictions from
these relationships have proven infallible in wonderfully diverse
applications; an example is the nature of electromagnetic radiation
and its propagation in vacuum at the constant speed of light.
Maxwell’s equations are shown here in both integral (a) and
differential (b) form:
Faraday’s law:

Ampère’s law and Maxwell’s correction:

Gauss’s law (I):

Gauss’s law (II):

Charge conservation and continuity of charge:

Except in this chapter, in the subsequent treatment of

ferrohydrodynamics, the free charge pf and electric displacement D
are assumed absent, so that Gauss’s law (I) will play no role and
Ampère’s law will be absent the Maxwell term. In addition, the
current Jf is absent in nonconducting ferrofluid, making all terms in
the charge-conservation equation zero.
Following standard convention, the polarization P and
magnetization M are given by the defining equations

It has already been shown, in Chapter 1, that the definition of M

permits the identification of μM with the dipolar moment density in
a substance. The field variables appearing in Maxwell’s equations
are defined in Table 3.1. L stands for a line or contour integral
around a closed loop L, and d/dt = ∂/∂t + v· ∇ is the substantial
derivative introduced in Chapter 1. The primed quantity E’ in
(3.22a), is the value measured by an observer moving with the
contour L at the point in question.

Of the field variables appearing in Table 3.1, H and B have already

been introduced. Magnetic fields are produced both by electric
currents and magnetically polarized matter. In matter it is electron
spin, which has no classical analog, that principally generates the
polarization. The free current density Jf results from the
translational motion of free electric charge having density πf. The
term “free” distinguishes such charge from the “bound” charge
associated with electrically polarized matter. The electric field
vector E at a point represents the force experienced by a test
electric charge located at the field point. The force is operationally
measurable when the field point is accessible to a probe, as in air,
and within matter the field is calculable by definition from the field
equations. The displacement field D is the electric analog of B, the
field of magnetic induction. When the polarization P is known
independently, D can be calculated from the defining equation
(3.27). For example, a permanently polarized rod or electret is the
analog of a permanent magnet, with P the analog of μ0M.

Integral equations
Integral equations are especially useful in moving boundary
problems and in deriving general expressions for interfacial or
boundary conditions. The integral equations as written are true for
stationary contours and surfaces. However, if the contours and
surfaces are moving, then the field quantities must be written in the
moving frame of reference (the primed frame). This in turn requires
transforming the field quantities from the moving frame to the
observer’s frame to obtain useful working relationships. The general
transformation of field variable is accomplished consistent with
special relativity theory with the assertion that Maxwell’s differntial
equations must exhibit the same form in both frames, an assertion
leading to the Lorentz transformation (see, e.g., Jackson 1975,
Chapter 11). For velocities small compared to the velocity of light
(i.e., ), the only numerically significant transformation to be
employed in the integral equations relates to the electric field E (see
below).
The line integral in the integral form of Faraday’s law is known as
the electromotive force (EMF). Faraday’s integral law states that the
EMF is induced by the changing flux of B through the circuit L, and
the flux can be changed by changing the magnetic induction or the
shape or orientation or position of the circuit. The time derivative
appearing on the right side of the law can be brought inside the
integral with the aid of the surface integral version of the Reynolds’
transport theorem (see Appendix 1):

However, because ∇·B = 0 and by Stokes’s theorem (see Appendix

1), (3.29) becomes

Using equation (3.22a), to eliminate the integral on the left side of

(3.30) gives

Combining the contour integral terms of equation (3.31a), then

yields the form

or

where
This is the Galilean transformation relating E to E′. It is an
approximation, although a very good one; through the use of
Reynolds’ theorem it has been tacitly assumed that x′ = x − vxt, y′ =
y,z′ = z, and t′ = t connects coordinates of space and time in a
primed system moving at speed vx along the x axis relative to an
unprimed system. The exact relationship between E and E′ is
developed in many electromagnetics texts. Return now to equation
(3.31b), and let Stokes’s theorem be applied to transform the left
side to an integral over surface area; noting the arbitrariness of the
surface of integration now leads at once to the differential form of
Faraday’s law, equation (3.22b). This equation, and in fact the
complete set of the differential Maxwell equations, are always
correct as written.

Differential equations
As already noted, a magnetic field is produced not only from
polarized matter but also from motion of charge, that is, from
electric current. The influence of the current density Jf appears as
the first term on the right side in the differential form of Ampère’s
law.
Although he was unable to prove it experimentally, Faraday
believed that a time-varying electric field should also generate a
magnetic field. It was left for Maxwell to show that Faraday was
right and that without this amendment Ampère’s law and
conservation of charge are inconsistent. Thus, if the divergence of
the incomplete differential form of Ampere’s law is taken, it is found
that

because ∇ · (∇ × A) = 0 for any vector field A with continuous partial

derivatives of second order. This result, however, is in contradiction
with (3.26b), if a time-varying charge is present. Maxwell realized
that if the displacement current ∂D/∂t is added to the right-hand side
of Ampère’s law, charge conservation is automatically satisfied.
Thus, if the divergence of (3.23b), is taken and the resulting
equation is simplified with the aid of (3.24b), the charge-
conservation equation, (3.26b), is obtained. Unlike in fluid
mechanics, where the mass conservation equation is an independent
relation, Maxwell’s equations have built into them the principle of
charge conservation.
Equation (3.24a), the first of Gauss’s laws, is a direct consequence
of Coulomb’s law for point electric charges. Equation (3.25a), states
that isolated magnetic poles do not exist, and to date none have
been found. The differential form was derived as (3.3). The
equivalence of the differential and integral forms of Gauss’s laws is
immediately demonstrated through use of the divergence theorem.
Note that Maxwell’s equations have more dependent variables
than the number of equations relating them. In working with
macroscopic aspects of electromagnetism it is convenient to relate
the field quantities by constitutive laws defining the functions ,
and σ. For isotropic but nonlinear materials these are

where is the permittivity, μ is the permeability, and σ is the

electrical conductivity of the material. In free space and μ are
constants having the values 0 = 8.854 × 10 −12 −1
F·m and μ0 = 4π ×
10−7 H·m−1. The speed of light in vacuum is c = ( 0μ0)−1 = 2.9979 π
108 m ·s−1. Equation (3.35) is Ohm’s law for moving media and is
correct only for systems moving at velocities much less than the
speed of light.
In most work to date in FHD the medium is ferromagnetically
responsive and both the free current density and Maxwell’s
displacement current are negligible. Hence the field equations of
FHD are usually employed in the magnetostatic limit of Maxwell’s
equation:

These relationships were previously developed as (3.3) and (3.4),

respectively. It is not difficult to imagine circumstances in which an
alternating current is present and the displacement current
produces a magnetic field or in which current flows in a conducting
ferrofluid, so that (3.23b), must be used in place of (3.37). Also, for
example, if appreciable amounts of free charge were present, then
the full description would necessarily include (3.22) and (3.24).
Moreover, in dealing with the sources of field it is important to
employ additional terms of Maxwell’s equations. Illustrations of
sources are provided with reference to the complete Ampère’s law in
the following examples.
Example - infinitely long straight conductor as illustration of
Ampère’s law in steady state, ∂D/∂t = 0 and Jf ≠ 0: A long, straight
wire of radius R carries a constant current I, as shown in Figure 3.8.
Determine the distribution of magnetic field H established by this
current.

3.8 Application of Ampère’s circuital law to a long current-carrying wire.

Solution: In this situation displacement current is absent because

the fields are time steady. Ampère’s law relates magnetic field to
sources of current; from (3.23a), the appropriate form is

From symmetry considerations based on the circular contour

indicated in Figure 3.8, the free current gives rise to an azimuthal
field H = Hϕ given by

or
Thus, the field increases linearly with radius within the wire and
dies off inversely with radius outside the wire.
Example - alternating current electrical capacitor as illustration of
the unsteady state Ampère’s law, Jf = 0 and ∂D/∂t ≠ 0: Determine the
magnetic field established within the gap of an air capacitor as
sketched in Figure 3.9. The metallic capacitor plates are electrically
driven with an alternating source of voltage V. Neglect fringe field
effects.

3.9 Application of Ampère’s circuital law to the magnetic field produced by a

displacement current.

Solution: The alternating voltage creates an alternating current i

in the conductors and an accumulation of free charge on the
capacitor plates. No charge current flows in the air gap, but a
uniform field D = 0E is established in the gap, and its time rate of
change is the displacement current ∂D/∂t. From Ampère’s circuital
law with current Jf = 0.
 The sketch indicates a portion of a stationary contour L of radius r
enclosing an area S in the gap space. The integrals can be evaluated
using symmetry considerations to give

where H is constant in circular loops about the center.

Since the applied voltage V = EL, where L is the gap length, the
field due to the displacement current is

3.5 Energy density of the electromagnetic field

When a magnetic field or an electric field is established in a
region of space, an expenditure of energy is necessary over and
above any energy consumed in irreversible processes such as ohmic
heating. This energy can be evaluated from knowledge of the spatial
distribution of the field vectors, independent of any detailed process
employed in reaching the distribution. Thus, an energy density can
be attributed to an electromagnetic field distribution. The result,
important in later developments, is derived in this section based on
use of the full set of Maxwell’s equations.
It will be supposed that fields are established in an arbitrary
system starting from a field free state. The fields are assumed to be
generated by electric current distribution Jf in electric field E over
the field volume V. Thus, in any differential volume of the system the
differential work done by external sources is given by

The term E·Jf can be manipulated into the following form, where
only electromagnetic field variables appear on the right side:
Jf was eliminated using Ampère’s law, (3.23b), in the first term on
the right side, and the second term is identically zero on account of
Faraday’s law, equation (3.22b). If use is now made of the vector
identity

(3.44) can be rewritten as an integral expression for the work done

on the system by external sources to establish the field:

The second term on the right side can be converted to a surface

integral with the aid of the divergence theorem:

where E × H is called the Poynting vector and has units of watts per
square meter. The Poynting vector formally describes the flux or flow
of electromagnetic energy through space; only its integral over a
closed surface has measurable significance. On the assumption that
H and E go to zero on the system boundaries, (3.47) becomes

The volume integrations are carried out over stationary volume

elements, so (3.49) becomes

where it is assumed that the fields are initially zero. The integrand is
energy density of the electromagnetic field in units of newton-meters
per cubic meter.
Based on the form given in (3.50), H·dB will be considered to
represent the differential contribution to local energy density of the
magnetic field. When H and B are collinear, as in soft magnetic
materials under equilibrium conditions, the term H·dB is expressible
in terms of the field magnitudes as HdB.

3.6 Transformed expression for the field energy

In the previous section (3.50) encompasses a general
expression for the magnetostatic energy of an entire system, which
could be composed, for example, of a volume of magnetic fluid
together with regions containing no magnetic material or other
regions containing a source of magnetic field - such as a current-
carrying coil, permanent magnets and permeable iron structure, or
other materials. However, at times it will be convenient to work with
an expression that isolates just the magnetostatic energy identified
with the magnetic fluid, and that is what is accomplished in this
section.
Consider Figure 3.10. Let V1 be a linear and isotropic medium of
permeability μ1, and let V2 be the volume occupied by magnetic
fluid, which will be treated as linear and isotropic with permeability
μ2. Thus, its energy density from (3.50) becomes H · B. The source
of magnetic field is a motionless loop of wire considered perfectly
conductive (i.e., of infinitely large electrical conductivity σ and
carrying an electric current I. From Faraday’s law in integral form,
(3.22a), it is seen that the total flux passing through the loop ϕ = ∫B
· dS is constant when the line integral of E′ evaluated around the
loop is zero. The loop is assumed to be perfectly conducting, so E′
must be zero everywhere in it; otherwise, from Ohm’s law, the
current density Jf = σE′ would become infinite. Hence the flux ϕ
remains constant when the surroundings are modified as by moving
in the volume of magnetic fluid. The essential point is that no
additional energy storage is possible in the current loop.
 3.10 Developing an expression for magnetostatic energy of a magnetizable fluid.

A one-step process can be considered in which the magnetic fluid

initially is located distantly outside the volume enclosed by surface S
and then is brought into this volume, thus displacing an equal
volume of the first medium. The energy change resulting from the
introduction of the magnetic fluid to the region is

where V = V1 + V2. The initial field is represented by B0 and H0 =

−∇ψ0, the final field by B and H = −∇ψ where ψ0 and ψ are the
potential functions. With the aid of the identity

and the divergence theorem, the first integral on the right side of
the final line of (3.51) may be written
Now the outer boundary of S may be allowed to recede to infinity so
as to include all space. The integrand of the surface integral goes to
zero faster than the surface area grows, so that the surface integral
disappears. It is also true that ∇·(B − B0) is zero throughout V, and
the volume integral vanishes. Thus, the right side of (3.53) is zero
and hence

The second integral on the right of (3.51) may written as the sum
over the subvolumes V1 and V2:

This is because in V1

which makes the integral over V1 identically zero. In V2, B = μ2H

and B0= μ1H0 so that

Therefore,
If μ1 = μ0then in V2

so that

and(3.57)becomes

Equation (3.59) expresses, in terms of the magnetization vector M,

the energy effect in introducing a volume of magnetic fluid into a
fixed- source static magnetic field in free space. It is remarkable in
that first, the field energy that originally was expressed as an
integral over all space is here expressed as an integral over just the
fluid volume and, second, it is the original field H0 and not the final
field H that figures into the expression. These features prove very
convenient in handling certain problems that arise; an example is
the problem of spacing in a magnetic fluid labyrinth, discussed in
Chapter 7.

Comments and supplemental references

We cannot, of course, come even close to giving a full picture of one
of the oldest and richest subjects of physics, classical
electromagnetic theory. This subject, along with classical and
quantum mechanics, forms the core of present-day training for
physicists and is of considerable importance in electrical
engineering curricula. Consequently there are numerous good books
that the interested reader can turn to for further study. Excellent,
modern introductions are found in the texts of
Zahn (1979)
which emphasizes problem solving, and
Reitz, Milford, and Christy (1979)
More advanced treatments are those of
Landau and Lifshitz (1960)
 Panofsky and Phillips (1962)
Jackson (1975)
and the somewhat older, but still classic, texts of
Jeans (1925)
Stratton (1941)
Also of relevance and great value are books on field theory and
potential theory. Two fine examples are
Moon and Spencer (1961)
Kellogg (1953)
A splendid graduate-level text that develops the concepts of
electrohydrodynamics in parallel with some aspects of magnetism is
that of
Melcher (1981)
For a development that is radically different from traditional
treatments, see the mathematically awesome but precise article by
Truesdell and Toupin (1960)
whose principal objective is “to isolate those aspects of the theory
which are independent of the assigned geometry of space-time from
those whose formulation and interpretation depend or imply a
particular space-time geometry.”
The derivation of the macroscopic Maxwell’s equations by suitably
averaging over aggregates of atoms dates back to Lorentz. For a
fresh look at the subject see
Russakoff (1970)