Homework 4 Answers, 95.

657 Electromagnetic Theory I

Dr. Christopher S. Baird, UMass Lowell
Jackson 2.13
(a) Two halves of a long hollow conducting cylinder of inner radius b are separated by small
lengthwise gaps on each side, and are kept at different potentials V
1
and V
2
. Show that the potential
inside is given by
1(j, )=
V
1
+V
2
2
+
V
1
V
2
n
tan
1
(
2bj
b
2
j
2
cos
)
where is measured from a plane perpendicular to the plane through the gap.
SOLUTION:
Due to the symmetry of the problem, it is apparent that the solution will be best expressed in cylindrical
coordinates. Additionally, because the solution will be independent of the z coordinate, the problem
reduces to the two dimensions of polar coordinates (j ,) . Because the problem contains no charge,
the problem simplifies down to solving the Laplace equation
2
1=0 in polar coordinates and
applying the boundary condition 1(j=b , )=V () where:
V ()=

V
1
if n/ 2>>3n/ 2
V
2
if n/ 23n/ 2

The Laplace equation in polar coordinates is:

1
j

j
(
j
1
j
)
+
1
j
2

2
1

2
=0
Separation of variables leads to the general solution:
1(j, )=(a
0
+b
0
ln j)( A
0
+B
0
)+

+ ,+0
(a
+
j
+
+b
+
j
+
)( A
+
e
i +
+B
+
e
i +
)
We desire a valid solution at the origin, which is only possible if b
0
=0 and b

=0 so that the solution

becomes:
1(j, )=A
0
+B
0
+

+,+0
j
+
( A
+
e
i +
+B
+
e
i +
)
We desire a single, valid solution over the full angular range, so the single-value requirement means
1(j, )=1(j ,+2n) . When we apply this, we get:
A
0
+B
0
+

+ ,+0
j
+
( A
+
e
i +
+B
+
e
i +
)=A
0
+B
0
(+2n)+

+, +0
j
+
( A
+
e
i +( +2n)
+B
+
e
i +(+2n)
)
Which leads to B
0
=0 and =n where n=1, 2, ... We now have:
1(j, )=A
0
+

n=1

j
n
(
A
n
e
i n
+B
n
e
i n
)
Now apply the last boundary condition 1(j=b , )=V ()
V ()=A
0
+

n=1

b
n
(
A
n
e
i n
+B
n
e
i n
) (Eq. 1)
Let us first find the A
0
term. Integrate both sides over the full angular sweep.

0
2n
V () d =

0
2n
A
0
d +

n=1

b
n
(
A
n

0
2n
e
i n
d +B
n

0
2n
e
i n
d
)

n/ 2
n/ 2
V
1
d +

n/ 2
3n/2
V
2
d =A
0
2n
V
1
n+V
2
n=A
0
2n
A
0
=
V
1
+V
2
2
Let us now find the A
n
coefficients. Multiply (Eq. 1) on both sides by e
i n'
and integrate over all
angles :

0
2n
V () e
i n'
d =A
0

0
2n
e
i n'
d +

n=1

b
n
(
A
n

0
2n
e
i( nn' )
d +B
n

0
2n
e
i ( n+n')
d
)
Use the orthonormality condition

0
2n
e
i ( kk ') x
dx=2n6
k , k '

0
2n
d V () e
i n
=2nb
n
A
n
A
n
=
1
2nb
n

0
2n
d V ()e
i n
Plug in the explicit form of the potential on the boundary which breaks the integral into two parts:
A
n
=
1
2nb
n
|
V
1

n/ 2
n/ 2
d e
i n
+V
2

n/ 2
3n/ 2
d e
i n

A
n
=
1
2nb
n
|
V
1
|
e
i n
i n

n/ 2
n/ 2
+V
2
|
e
i n
i n

n/2
3n/ 2

A
n
=
(1)
( n+1)/ 2
nnb
n
|
V
1
V
2
and
A
n
=0 if n=even
Let us now solve for the B
n
coefficients. Take (Eq. 1) again and this time multiply by e
i n'
and
integrate over all :

0
2n
V () e
i n'
d =A
0

0
2n
e
i n'
d +

n=1

b
n
(
A
n

0
2n
e
i (n+n')
d +B
n

0
2n
e
i ( n'n)
d
)
Use the orthonormality condition

0
2n
e
i ( kk ') x
dx=2n6
k , k '

0
2n
V () e
i n
d =b
n
B
n
2n
B
n
=
1
2nb
n

0
2n
V ()e
i n
d
Plug in the explicit form of the potential on the boundary which breaks the integral into two parts:
B
n
=
1
2nb
n
|
V
1

n/ 2
n/ 2
d e
i n
+V
2

n/ 2
3n/ 2
d e
i n

B
n
=
1
2nb
n
|
V
1
|
e
i n
i n

n/ 2
n/ 2
+V
2
|
e
i n
i n

n/2
3n/2

B
n
=
(1)
( n+1)/ 2
nnb
n
|
V
1
V
2
and
B
n
=0 if n=even
Now that we have found all of the coefficients, the solution is determined:
1(j, )=
V
1
+V
2
2
+

n=1, odd

j
n (1)
(n+1) / 2
nnb
n
|
V
1
V
2

(e
i n
+e
i n
)
1(j, )=
V
1
+V
2
2
+
V
1
V
2
2
4
n

n=1, odd

(1)
(n+1)/ 2
j
n
nb
n
cos(n)
1(j, )=
V
1
+V
2
2
+
V
1
V
2
2
4
n


n=1, odd

(1)(1)
( n+1)/ 2
n
(
je
i
b
)
n
Now we recognize the Taylor expansion of the arctan: tan
1
( x)=

n=1, odd

(1)(1)
( n+1)/ 2
n
x
n
1(j, )=
V
1
+V
2
2
+
V
1
V
2
2
4
n

|
tan
1
(
je
i
b
)
Using the identity
tan
1
( z)=
i
2
| ln(1i z)ln (1+i z) and expanding the complex number z in its
components, we can prove the identity:
| tan
1
( z)=
1
2
tan
1
(
2 ( z)
1z
2
)
We now use this identity:
1(j, )=
V
1
+V
2
2
+
V
1
V
2
n
tan
1
(
2bj
b
2
j
2
cos
)
(b) Calculate the surface-charge density on each half of the cylinder.
SOLUTION:
As derived earlier using a Gaussian pillbox surface, the surface-charge density on a conductor is related
to the potential according to:
c=
|
c
0
d 1
dn

n=a
We assume that the problem is seeking the charge density on the inside surface of the conductor
because that is where we know the potential. In that case, the normal to the inside surface of the
conductor points in the opposite direction as the radial dimension, so that n=-.
c=
|
c
0
d 1
d j

j=b
c=
|
c
0
d
d j(
V
1
+V
2
2
+
V
1
V
2
n
tan
1
(
2bj
b
2
j
2
cos
))

j=b
c=
|
c
0
V
1
V
2
n
(
1
1+
(
2bj
b
2
j
2
cos
)
2
)
d
d j
(
2bj
b
2
j
2
cos
)

j=b
c=
|
c
0
V
1
V
2
n (
2bcos ( b
2
+j
2
)
(b
2
j
2
)
2
+( 2bjcos)
2
)

j=b
c=c
0
V
1
V
2
nbcos
Jackson 2.23
A hollow cube has conducting walls defined by six planes x = 0, y = 0, z = 0, and x = a, y = a, z = a.
The walls z = 0 and z = a are held at constant potential V. The other four sides are at zero potential.
(a) Find the potential (x, y, z) at any point inside the cube.
SOLUTION
The problem contains no charge, so the electric potential is described everywhere inside the cube by
the Laplace equation:

2
1=0
This problem has a boundary, the cube, that best matches rectangular coordinates. In rectangular
coordinates, the Laplace equation becomes:

2
1
x
2
+

2
1
y
2
+

2
1
z
2
=0
Using the method of separation of variables, the most general solution to this equation is:
1( x , y , z)=( A
o0
+B
o0
x)( A
0
+B
0
y)( A
0
+B
0
z)
+

( A
o0
+B
o0
x)( A

e
i y
+B

e
i y
)( A

e
z
+B

e
z
)
+

o
( A
o
e
i o x
+B
o
e
i o x
)( A
0
+B
0
y)( A

e
z
+B

e
z
)
+

o,
( A
o
e
i ox
+B
o
e
i o x
)( A

e
i y
+B

e
i y
)( A

e
z
+B

e
z
)
Apply the boundary condition, (x=0, y, z)=0
0=( A
o0
)( A
0
+B
0
y)( A
0
+B
0
z)
+

( A
o0
)( A

e
i y
+B

e
i y
)( A

e
z
+B

e
z
)
+

o
( A
o
+B
o
)( A
0
+B
0
y)( A

e
z
+B

e
z
)
+

o ,
( A
o
+B
o
)( A

e
i y
+B

e
i y
)( A

e
z
+B

e
z
)
This must be true for all y and z, so that each term must vanish separately.
This forces A
0
=0 and B

=-A

. The solution now becomes:

1( x , y , z)=( B
o0
x)( A
0
+B
0
y)( A
0
+B
0
z)
+

( B
o0
x)( A

e
i y
+B

e
i y
)( A

e
z
+B

e
z
)
+

o
A
o
sin(o x)( A
0
+B
0
y)( A

e
z
+B

e
z
)
+

o ,
A
o
sin(o x)( A

e
i y
+B

e
i y
)( A

e
z
+B

e
z
)
Similarly, the boundary condition, (x, y=0, z)=0 leads to A
0
=0 and B

=-A

which gives the solution:
1( x , y , z)=( B
o0
x)( B
0
y)( A
0
+B
0
z)
+

( B
o0
x) A

sin( y)( A

e
z
+B

e
z
)
+

o
A
o
sin (ox)( B
0
y)( A

e
z
+B

e
z
)
+

o,
A
o
sin(o x) A

sin( y)( A

e
z
+B

e
z
)
Now apply the boundary condition (x=a, y, z)=0
0=( B
o0
a)( B
0
y)( A
0
+B
0
z)
+

( B
o0
a) A

sin( y)( A

e
z
+B

e
z
)
+

o
A
o
sin(oa)( B
0
y)( A

e
z
+B

e
z
)
+

o ,
A
o
sin(oa) A

sin ( y)( A

e
z
+B

e
z
)
The only way this can be true for all y and z is if B
0
=0 and =n/a where n = 0, 1, 2,...
1( x , y , z)=

n
A
n
sin
(
nnx
a
)
( B
0
y)( A

e
z
+B

e
z
)+

n ,
A
n
sin
(
nnx
a
)
A

sin ( y)( A

e
z
+B

e
z
)
Similarly, apply the boundary condition (x, y=a, z)=0 to get B
0
=0 and =m/a where n = 0, 1, 2,...
1( x , y , z)=

n, m
A
n
sin
(
nnx
a
)
A
m
sin
(
mn y
a
)
( A

e
z
+B

e
z
)
By definition we have
2
=o
2
+
2
so that we now know =.(n
2
+m
2
) n
2
/ a
2
. We can combine
several constants so that we now have:
1( x , y , z)=

n, m
sin
(
nnx
a
)
sin
(
mny
a
)
( A
n ,m
e
z .(n
2
+m
2
) n
2
/ a
2
+B
n ,m
e
z .(n
2
+m
2
) n
2
/a
2
)
Now apply the boundary condition (x, y, z=0)=V
V =

n ,m
sin
(
nnx
a
)
sin
(
mny
a
)
( A
n, m
+B
n, m
)
Multiply both side by sin
(
n' nx
a
)
sin
(
m' n y
a
)
and integrate over x and y from 0 to a

0
a

0
a
V sin
(
n' nx
a
)
sin
(
m' n y
a
)
dx dy
=

n, m

0
a
sin
(
nnx
a
)
sin
(
n' nx
a
)
dx

0
a
sin
(
mny
a
)
sin
(
m' ny
a
)
dy( A
n, m
+B
n, m
)
Due to orthogonality, each integral on the right is zero, except when n=n' and m=m'

0
a

0
a
V sin
(
nnx
a
)
sin
(
mny
a
)
dx dy=
a
2
4
( A
n, m
+B
n, m
)
( A
n ,m
+B
n , m
)=
4V
a
2

0
a
sin
(
nnx
a
)
dx

0
a
sin
(
mny
a
)
dy
( A
n ,m
+B
n , m
)=
4V
a
2
a
nn
|1(1)
n

a
mn
| 1(1)
m

A
n , m
+B
n, m
=
16V
n mn
2
n, m = odd

Now apply the final boundary condition (x, y, z=a)=V
V =

n ,m
sin
(
nnx
a
)
sin
(
mny
a
)
( A
n, m
e
a.(n
2
+m
2
) n
2
/a
2
+B
n , m
e
a .( n
2
+m
2
) n
2
/ a
2
)
Repeat the process done above to get
A
n , m
e
.( n
2
+m
2
)n
2
+B
n ,m
e
.(n
2
+m
2
) n
2
=
16V
n mn
2
n, m = odd
Solve the system of equations in the boxes above to find:
A
n , m
=
8V
nmn
2
e

1
2
.( n
2
+m
2
) n
2
cosh ((n/ 2) .n
2
+m
2
)
B
n, m
=
8V
nmn
2
e
1
2
.(n
2
+m
2
) n
2
cosh((n/ 2) .n
2
+m
2
)
The final solution is now:
1( x , y , z)=

n ,m odd
sin
(
nnx
a
)
sin
(
mny
a
)
8V
n mn
2
(
e

1
2
.(n
2
+m
2
) n
2
e
z .( n
2
+m
2
) n
2
/ a
2
+e
1
2
.( n
2
+m
2
) n
2
e
z .(n
2
+m
2
) n
2
/a
2
cosh((n/ 2) .n
2
+m
2
)
)
1( x , y , z)=

n ,m odd
16V
n mn
2
sin
(
nnx
a
)
sin
(
mny
a
)
(
cosh ((n/ 2) .n
2
+m
2
((2 z/ a)1))
cosh ((n/ 2) .n
2
+m
2
)
)
(b) Evaluate the potential at the center of the cube numerically, accurate to three significant figures.
How many terms in the series is it necessary to keep in order to attain this accuracy? Compare your
numerical result with the average value of the potential on the walls. See Problem 2.28.
SOLUTION:
The potential at the center of the cube is:
1(a/ 2, a/ 2, a/ 2)=
16V
n
2

n ,m odd
1
n m
sin
(
nn
2
)
sin
(
mn
2
)
(
1
cosh((n/ 2) .n
2
+m
2
)
)
1(a/ 2, a/ 2, a/ 2)=V |
16
n
2
sin
(
n
2
)
sin
(
n
2
)(
1
cosh ((n/ 2))
)

+
16
n
2
1
3
sin
(
n
2
)
sin
(
3n
2
)(
1
cosh((n/ 2) .10)
)

+
16
n
2
1
3
sin
(
3n
2
)
sin
(
n
2
)(
1
cosh((n/ 2) .10)
)

+
16
n
2
1
5
sin
(
n
2
)
sin
(
5n
2
)(
1
cosh ((n/ 2). 26)
)
+...

1(a/ 2, a/ 2, a/ 2)=V |0.34760.00750.0075+0.0002+...
Here is solution, to three significant figures, if you only keep the first term, or the first two terms, etc:
1
1 term
(a/ 2, a/ 2, a/ 2)=V |0.348
1
2 terms
( a/ 2, a/ 2, a/ 2)=V | 0.340
1
3 terms
( a/ 2, a/ 2, a/ 2)=V | 0.333
1
4 terms
( a/ 2, a/ 2, a/ 2)=V | 0.333
We only have to keep the first three terms to have the answer accurate to three significant figures. It is
obvious that the solution is converging to:
1(a/ 2, a/ 2, a/ 2)=
V
3
There are six walls on the cube and two sides have a non-zero potential V, so the average value of the
potential on the sides of the cube is
1
ave on surf
=
2V
6
=
V
3
. This leads to the interesting conclusion
that:
1(a/ 2, a/ 2, a/ 2)=1
ave on surf
(c) Find the surface-charge density on the surface z = a.
SOLUTION
c=
|
c
0
d 1
dn

n=a
We have solved the potential on the inside of the cube, so we can only use that potential to find the
surface charge density on the inside of the z = a surface. The normal to the inside surface is in the
negative z direction so that:
c=
|
c
0
d 1
dz

z=a
c=
|
c
0
d
dz

n, m odd
16V
nmn
2
sin
(
nnx
a
)
sin
(
mn y
a
)
(
cosh((n/ 2). n
2
+m
2
(( 2 z/ a)1))
cosh((n/ 2).n
2
+m
2
)
)

z =a
c=
|
c
0

n ,m odd
16V
n mn
2
sin
(
nnx
a
)
sin
(
mny
a
)
(
sinh((n/ 2).n
2
+m
2
((2 z / a)1))
cosh((n/ 2).n
2
+m
2
)
)
(n/ a) .n
2
+m
2

z=a
c=
16c
0
V
na

n ,m odd
.n
2
+m
2
nmn
sin
(
nnx
a
)
sin
(
mn y
a
)
(tanh ((n/ 2).n
2
+m
2
))
Jackson 2.26
The two-dimensional region, a, 0 , is bounded by conducting surfaces at = 0, = a, and
= held at zero potential, as indicated in the sketch. At large the potential is determined by some
configuration of charges and/or conductors at fixed potentials.
a

(a) Write down a solution for the potential 1(j, ) that satisfies the boundary conditions for finite .
SOLUTION:
We can think of the charges away from this rounded corner as external to the problem, so that they
simply create some boundary condition on the potential at large . The region near the corner has no
charges and is described by the Laplace equation.

2
1=0
In polar coordinates, the Laplace equation becomes:
1
j

j
(
j
1
j
)
+
1
j
2

2
1

2
=0
Using the method of separation of variables, the general solution is found to be
1(j, )=(a
0
+b
0
ln j)( A
0
+B
0
)+

+ ,+0
(a
+
j
+
+b
+
j
+
)( A
+
e
i +
+B
+
e
i +
)
Apply the boundary condition 1(j, =0)=0
0=( a
0
+b
0
ln j)( A
0
)+

+,+0
( a
+
j
+
+b
+
j
+
)( A
+
+B
+
)
To hold true for all values of we must have
A
0
=0
and B
+
=A
+
. The solution now becomes:
1(j, )=(a
0
+b
0
ln j)( B
0
)+

+,+0
(a
+
j
+
+b
+
j
+
) A
+
sin(+)
Apply the boundary condition 1(j, =)=0
0=( a
0
+b
0
ln j)( B
0
)+

+,+0
( a
+
j
+
+b
+
j
+
) A
+
sin(+)
To hold true for all values of we must have
B
0
=0
and +=
nn

where n=1, 2, 3... which gives

1(j, )=

n=1

(
a
n
j
nn/
+b
n
j
nn/
)
A
n
sin
(
nn

)
Apply the boundary condition 1(j=a ,)=0
0=

n=1

(
a
n
a
nn/
+b
n
a
nn/
)
A
n
sin
(
nn

)
To hold true for all values of angles we must have 0=a
n
a
nn/
+b
n
a
nn/
which leads to:
b
n
=a
n
a
2nn/
The solution at this point takes the form (where several constant factors have been combined with the
last remaining undetermined constant):
1(j, )=

n=1

A
n
((
j
a
)
nn/

(
j
a
)
nn/
)
sin
(
nn

)
(b) Keeping only the lowest nonvanishing terms, calculate the electric field components E

and E

and
also the surface-charge densities (, 0), (, ), and (a, ) on the three boundary surfaces.
SOLUTION
The electric field defined in terms of the electric potential is:
E=1
In polar coordinates this becomes:
E= j
1
j

1
j
1

or presented differently:
E
j
=
1
j
and
E

=
1
j
1

E
j
=

j

n=1

A
n
((
j
a
)
nn/

(
j
a
)
nn/
)
sin
(
nn

)
E
j
=

n=1

A
n
nn
a
((
j
a
)
nn/1
+
(
j
a
)
nn/ 1
)
sin
(
nn

)
The lowest non-vanishing term is:
E
j
=A
1
n
a
((
j
a
)
n/ 1
+
(
j
a
)
n/ 1
)
sin
(
n

)
The other component can also be easily done:
E

=
1
j

n=1

A
n
((
j
a
)
nn/

(
j
a
)
nn/
)
sin
(
nn

)
E

n=1

A
n
nn
a
((
j
a
)
nn/ 1

(
j
a
)
nn/1
)
cos
(
nn

)
The lowest non-vanishing term is:
E

=A
1
n
a
((
j
a
)
n/ 1

(
j
a
)
n/1
)
cos
(
n

)
The total electric field with all the terms is then:
E= j E
j
+

E=

n=1

A
n
nn
a |
j
((
j
a
)
nn/1
+
(
j
a
)
nn/ 1
)
sin
(
nn

((
j
a
)
nn/ 1

(
j
a
)
nn/1
)
cos
(
nn

The surface charge density (, 0) is found using the pillbox Gaussian surface, which yields:
c( j, 0)=
|
c
0
En

n=n
0
c( j, 0)=
|
c
0
E

=0
c( j, 0)=
|
c
0
E

=0
c( j, 0)=
|
A
1
nc
0
a
((
j
a
)
n/ 1

(
j
a
)
n/1
)
cos
(
n

=0
c( j, 0)=A
1
nc
0
a
((
j
a
)
n/ 1

(
j
a
)
n/1
)
The surface charge density (, ) is found in a similar manner:
c( j, )=
|
c
0
E(

=
c( j, )=
|
c
0
E

=
c( j, )=
|
A
1
nc
0
a
((
j
a
)
n/1

(
j
a
)
n/ 1
)
cos
(
n

=
c( j, )=A
1
nc
0
a
((
j
a
)
n/ 1

(
j
a
)
n/ 1
)
The charge densities on the two flat surfaces are equal, which is what we would expect because of the
symmetry of the problem.
The surface charge density (a, ) obeys:
c( a , )=
|
c
0
E j

j=a
c( a , )=
|
c
0
E
j
j=a
c( a , )=
|
A
1
nc
0
a
((
j
a
)
n/ 1
+
(
j
a
)
n/1
)
sin
(
n

j=a
c( a , )=A
1
2nc
0
a
sin
(
n

)
(c) Consider = (a plane conductor with a half-cylinder of radius a on it). Show that far from the
half-cylinder, the lowest order terms of part b give a uniform electric field normal to the plane. Sketch
the charge density on and in the neighborhood of the half-cylinder. For fixed electric field strength far
from the plane, show that the total charge on the half-cylinder (actually charge per unit length in the z
direction) is twice as large as would reside on a strip of width 2a in its absence. Show that the extra
portion is drawn from regions of the plane nearby, so that the total charge on a strip of width large
compared to a is the same whether the half-cylinder is there or not.
SOLUTION
If = , the solution for the total electric field reduces to:
E=

n=1

A
n
n
a |
j
((
j
a
)
n1
+
(
j
a
)
n1
)
sin ( n)

((
j
a
)
n1

(
j
a
)
n1
)
cos ( n)

the lowest order term is:

E=A
1
1
a
|
j
(
1+
(
a
j
)
2
)
sin ( )

(
1
(
a
j
)
2
)
cos ()

Far away from the half-cylinder, >> a, and thus 1 >> (a/) and 1 >> (a/)
2
so that
E=A
1
1
a
| jsin ( )+

cos()
E=
A
1
a

j
This is just a uniform electric field in the y direction, normal to the conducting plane.
The charge density on the half cylinder has the form
c( a , )=c
0
sin() where c
0
=A
1
2c
0
a
and on the sides:
c( j, 0)=c(j, )=
c
0
2
(
1
(
j
a
)
2
)
The total charge on the half-cylinder (per unit length in the z direction) is
Q
half-cyl
=A
1
2c
0
a

0
n
sin ( ) ad
Q
half-cyl
=A
1
4c
0
If the same electric field were used ( E=A
1
/ a

j as derived above) and the cylinder were replaced

with a strip 2a wide, the charge density on the strip would be
c( y=0)=
|
c
0
Ej

y=0
c( y=0)=c
0
A
1
/ a
The total charge (per unit length in the z direction) on the strip is:
Q
strip
=c2a
Q
strip
=A
1
2c
0
So that
Q
strip
=
1
2
Q
half-cyl
Consider a larger strip with width l that includes the central region already focused on. The total charge
with the half-cylinder included is:
Q
1
=2

a
l
(A
1
)
c
0
a
(
1
(
j
a
)
2
)
d j+Q
half-cyl
Q
1
=2( A
1
)
c
0
a
(l a)2( A
1
)a c
0
(1/ l1/ a)+Q
half-cyl
Q
1
=2 A
1
c
0
|
l
a

a
l

For l >> a :
Q
1
=
2l c
0
A
1
a
The total charge without the cylinder is
Q
2
=c2l
Q
2
=
|
c
0
E
y

y=0
c 2l
Q
2
=(c
0
A
1
/ a) 2l
Q
2
=
2l c
0
A
1
a
It now becomes apparent that
Q
1
=Q
2
. The total charge over a large region is the same with and
without the half-cylinder. This means that the extra charge on the half-cylinder is drawn from the
regions nearby on the planes.