56

Mathematical Ph .
Ys rcJ

2
Thus, dS = dx2 + dy2 + dz2
2 2
= a sinh 2 u cos 2 v(d") 2 + a 2 cos h2 11 sin v(dv)1
2 2
+ a 2 cosh 2 u sin 2 v(du )2 + a 2 sinh 2 u cos v(dv)2 + dz
2 2
= a 2(sinh 2 u + sin 2 v ) d,l + a2{sinh 2 11 + sin v ) rlv + di.
Thu s fo r lhi s system
2 2 7
h 1 = hu =a(sinh 2 u + sin 2v) 112 • /z 2 = hu = a( sinh u + sin v)" h3 =hz = J.
(0 Parabolic Cylindrical Co-ordinates (~, 17, z) . In thi s case the transformation eq uations are
2
y = l12(17 2 - ~ ) . z =z
x= ~T). . . . (1·105)
which generate two sets of orthogonal parabolic cylinders for
~ = const., -QO < ~ < oo and 17 = const. , 0 ~ 17 < 00 , respectively.
From eq uat ions ( 1-105), we have
dx = d~11 + ~drJ , dy = 11 d11 - ~ d~, dz = dz,
which gi ves
and hence fo r th is system
hi = h~ = (~2 + 11 2)"2, '12 = hri = (~2 + 112)112, h3 = h, = I.
§ 1-22. The Vector Integration. If there is a vector V = V(t) , where t is a scalar variable. then we
define the vector integration as
f V(t) dt = f [Vi:(t)i + Vy(t)j + Vz(t) k] dt
= if Vx dt + j f Vy dt + k f Vz dt, .. . (l -106)
where i, j and k are the unit vectors in three mutually perpendicular directions, taken as axes of reference.
The most important integrals encountered in Vector Analysis are the line integral, the surface integral
and the volume integral.
V
(a) The Line Integral. The integration of a vector along a curve is z
called line integral.
Let us consider a vector field defined by
V = V(x, y, z).
In th is field, take a curve C1, end points of which are P 1 and P 2. At any p
point A of this curve the value of vector is V. >----- - Y
Consider a small element of length l>r at A . Then V · dr is the scalar 0
product of the length of element and component of V in its direction, i.e.,
tangentially to the curve. Thus
X
V · dr = V cos 0 dr Fig. 1-43
If V varies in magnitude and direction from point to point along the curve,
then the integral
f - V cos 0 d r
f P2
P1
V · dr =
P,

P1
is defined as the line integral ofV along the curve c 1•
f
If c V · dr is independent of the path, then the vector field V(x, y , z) is said to be conservative or
non-curl o~ lamellar vector as discussed in § l · 16. In this case the line integral depends only on th e eo<l
value of the curve; so
s
57
Vector Analysis

Jc. V . dr =( 1

pl
V • dr =- J V • dr =-
C2
rP1
I

V · dr,

where C2 is another curve having the same end points P 1 and PJ. as those of C 1•

Hence J~ V · dr + J~ V · dr =p V · dr =(~ V · dr +('~ V · dr =0, 2 . . . (1-107)

where the sign f denotes integral round a closed curve.

Vector fields which satisfy the equation ( 1-107) are derived from the scalar field by taking the gradient
of the latter and can be written as V = Vq> (as discussed in § 1-16). The curl of such a vector field is zero.
The line integrals are of very frequent occurrence. For example, if V is force and dr is an element of
the path of a particle along any curve, then the line integral denotes the work done in displacing the particle
from one end point to the other end point. Again if V is the electric field strength, then the line integral
expresses the potential difference between the end points.
Ex. 1. If V = (3x2 + 6y) i - 14yz j + 20xz2 k, then evaluate JC V · dr, Z
(1, 1, 1)
where C is a straight line joining (0, 0, 0) and ( 1, 1, 1).
Let the parametric equations to the straight line be
X =t, y =t, Z= t,
so that V = (3r2 + 6t) i - 14t2 j + 20t3 k r------Y

and r = t i + t j + t k.
dr = dt i + dt j + dt k.

Hence f V · dr =r
C
=
t =0
1
(6t - 11? + 20f) dt
- ,
X
Fig. 1·44
I
= 13? --3 U 3
t + 5t
4
I 0
=3--3
11
+ 5 =-.
13
3
Ans.

Ex. 2. If F =(5.xy - 6x2 ) i + (2y - 4x) j, find J F · dr along the curve C given by y =x
C
3
in x-y plane,

from the point (1, 1) to (2, 8).

Along the curve y = x3, we have
F = (5x4 - 6x2) i + (2x3 - 4x) j and r =x i + y j =x i + x3 j.
dr = dx i + 3x2 dx j,

so that
f F . dr = r {(5x
C I
4
- 6x2) i + (2x3 - 4x) j} · (i + 3x2 j) dx

= r{I
(5x4 - 6x2 + (2x
3
- 4x)3x2 } dx

= f {6x + 5x + 12.x
2

I
5 4 3
- 6x
2
} dx

= I x6 + x5 - 3x4 - 2x3 I:
=(64-1)+(32-1)-2(8-1)-3(16-1)
Ans.
=63 + 31 - 45 - 14 =35.
(b) The Surface Integral. Let us consider a vector field V = V(x, y, z) and draw a surface Sin this
field. Let us now take an element of area 8S on the surface.
 58
Mathematica[ Pl1 .
Ysics
Draw the positive normal n of unit length on this element. The positive normal to
the closed surface is drawn outward from the enclosed volume, while for open surface n
this is in the same direction as that of the vector concerned. Let 0 be the angle between
the directions of n and V . Then the component of V perpendicular to 8S is

V · n = V cos 0 lnl = 1.
The flux of V through the element is V · n dS. The integral of this flux taken over
the surface is called the Total Flux or Surface Integral of V through the whole surface
S, i.e. ,

HV . n dS = HV cos edS. .. . (1-108)

Fig. 1·45
s s
= H(Vx dSx + Vy dSy + Vz dSz)
s
Note. The notation # is sometimes used to denote the integration over the closed surface S.
s
Examples of surface integration may be considered as follows:
Suppose V denotes the vector velocity of a moving fluid in which a fixed surface S is drawn. At any
point V · n dS denotes the amount of fluid passing normally through the surface element dS in unit time.
The integral of this normal contribution expresses the flow of fluid through the whole surface in unit time.
Similar ideas apply to other fluxes, e.g ., of electric or magnetic induction.
Ex. 1. Evaluate JJ V • n dS, where V = 4xzi -y2j + yzk and Sis the surface of the cube bounded by
x=0, x= 1, y=0, y= 1, z=0 and z= 1 planes.
Referred to the adjoining figure, we have
z
For face DEFG, x = 1, n = i, so that

·
JJ V-nds=J
I

0 0
f (4zxi-y2j+yzk)·idydz
I

DEFG
I I
=f f (4zi-y2j+yzk)·idydz (sincex=l)

;1 l
0 0

= (f >, dydz= ( 4 [ dy=+ =2.

X
For face ABCO, x = 0 and n = -i. Fig. 1-46

fl
ff V. n dS = _ ( ( (-y2j + yzk) · i dy dz= 0,
ABCO
)
For face ABEF, y = 1, n = j.
ff V . n dS = f'f' [4zx i - j + z k] · j dy dz
0 0
ABEF

For face CDGO, Y = 0, "= -j.

ff V. n dS =foIfo -4zxi . j dx dz= 0

I

CDGO
 59
Vector Analysis

For face BCDE, z = 1, n = k.

JJ V · n dS = ( ( (4xi - l j + yk) · k dy dx
BCDE O O

For face AOGF, z = 0, n = -k.

JJ V · n dS = - ( ( -y\j · k dx dy = 0 (as j · k = 0)
AOGF O O ,

Hence
JI V · n dS = JI V • n dS + JI V · n dS + JI V · n dS
DEFG ABCO ABEF

+ JI V · n dS + JJ V · n dS + JJ V · n dS
CDGO BCDE AOGF

1 3
=2+0-1+0+2+0=2· Ans.

Ex. 2. Evaluate JI V · dS over the faces of a cube given by

0$x,y,z$l and V=x2 i+y2j+z2k.
Referred to the figure etc., of the last example, we have
ff V · dS = 1 , JI V · dS =0 ;
DEFG OABC

HV -dS= 1, HV · dS=0;
ABEF OCDG

JI V · dS= 1, HV-dS=0;
EBCD OAFG

Hence JJ V · aS = 1 + 0 + 1 + 0 + 1 + 0 = 3. Ans.
Ex. 3. Evaluate the surface integral of a vector V =xi+ y j + z k over the closed surface of a cylinder
with the circular base of radius a in x-y plane and length h along z-axis. z
For the calculation of surface integral we need the surface normal n. On
a
the top surface of the cylinder shown in Fig. 1•47,
n=k and V · n = V · k = z =h.
h
Thus ff V · dS = fJ V · n ds = h fJ ds = hna 2

on top surface.
On the bottom surface of the cylinder,
n = -k and V · n = -V · k = - z = 0 y
(since for the bottom surface z = O). X

Thus JJ V · dS =fJ V · n ds = 0 Fig . 1-47

On bottom
surface
60 Mathematical Ph .
Ystcs

V, • norma 1 to the curved surface of the cylinder and

ec tor x 1• + y J• 1s ,l- therefore for this surface we have
2
x i +yj i x +j y and V-n=---2:.2:'._=a.
n ✓.l-+ y2 - a a
Thus fJ V . n ds = a ff ds = a (area of curved surface)
Curved
surface
= 21uih
The value of# V • n ds over the whole surface of cylinder is then
1ta.2h + 21uih = 31ui2h.
Ex. 4. Given V = 4yi + xj + 2zk, find ff (v' x V) • dS over the hemisphere x 2
+ y2 + z2 = a 2, z ~ 0.
Here curl V = -3k.
The surface of hemisphere with centre at origin is normal to r =xi+ yj + zk. Thus for this surface
n=.!.=~=xi+yj+zk r 3z
and hence (v' x V) · n = -3k · - = - - .
r a a a a

Thus JJ
(v' x V) · dS = JJ
(v' x V) · n ds' = -
3
a
z ds. ff
s
In spherical coordinates we have
z = r cos 8, ds = r2 sin 8 d8 dq>. a
For the surface in question r = a, then the required integral is
.f 2n fCOS 8
n/2
-3a - - a 2 sin 8 d8 d<!> = -3a 2
2n
d<!>
7112 J f
sin 8 cos 9 d0 = -3na 2.
q,=0 q, =O a q,=0 q,=0
A Theorem on Surface Integration. If R is the projection on the xy-plane of the surface S, then

Jf A· dS = Jf A · n ldx dkyl
S R D ·
.. . (1-109)

where n is unit normal on S and k is unit normal along z-axis.

Proof. Let us consider a surface S and take an infinitesimal small
element tiSP of this surface.
Let this surface be inclined to xy-plane at an angle 0. Then
ff A · dS is the limit of the sum
s
N
r, AP • n /:iSp as N➔ 00 and tiSP ➔ 0,
p=I
where Aµ is the value of the vector at the middle point of the smalll element X
t:,.YP
l:iSµ .
Fig. 1-48
The projection of tiSP on xy-plane is &pl:iYµ, so that
&~yp = tiSP cos 9 = tisp n · k.
where k is the unit normal for the surface element l:ixµl:!,.Yp,

Hence ff A
/:ixpl:iYp
tiSP = In· kj

5 dS= Lim
N
I:AP · nASP
"\G: dx dy
Fig. 1·49
S N ➔ oo p=I
f<>S,, ➔ 0
Vector Analysis
61

.
; :": LAP . n
6x2 6 yp __
In . kl -
ff A . n Incit .dykl
R
tu, ~ 0
6yr~ o

IUustration. Use this theorem to evaluate fJ (v' x V) . n dS , where V = y i + (x - 2xz)j - xy k and S

s
1s the surface of the sphere x2 + y2 + :. 2 = a 2 about the xy-planc.
Here, the value of c url V i.e . (V x V)
i j k
a
= ax a a
ay az = Xj + y j - 2z k.
y x- 2xz - xy

The normal to the surface x2 + J + -l- = a 2 is given by

grad S =v(x2 + y2 + ?-) =2xi + 2yj + 2zk,
. 2x i + 2y j + 2z k 2x i + 2y j + 2z k xi + y j + z k
so that unit normal n = . . = -;===c==~===
= ----
l2x1+ 2yJ + 2z kl ✓(4x2 + 4y2 + 4-l-) a
and k is the unit vector, normal to xy-plane .

Now ff (V x V) · n dS = ff (V x V) · n ~ .d~
. .
= ff (x + J - 2z k) ·
1 y
(x i+ yj +z k )dxdy
a za
R I

= ffx2 + y2 - 2z2 . dx dy = fJ x2 + l - 2(a 2 - x2 - y2) . d.x dy

R a z/a R ✓(a2-x2-J)
3 2
= ff (x22 + i>- 2a . dx Uy (put x = p cos cp, y = p sin cp, so that
I< ✓a -x2 - l
x2 + y2 = p 2 and dx dy = p dp dcp )

= R
3p2 - 2a2
IJ ✓(a1 - p2) . p dp dq,- R
- ff l 3(p2 - a2)
✓(a1 - p2) + ✓(a2 - p2)
a2 }
p dp d~

= ff l-3p ✓(a2 - p2) + ✓aa2p p

R
2
-
2
l r
dp dcp =
fl = 0
J27l
♦=0
- 3p ✓a2 - p2 dp dcp
+ rIoo
211
_k_dp
✓a 2 - p-y
d.p

= f 211 fn - 3p ✓a 2 - 2
r-cr-.,.-
p d~ dp +
f 2n Ja a2p
✓ d~ dp
~ .. o p = o o o a2 _ p2
(limits being independent, the change of order of integration is valid)

=('' [<a' -p' )J/2 I ti$ - a'C[✓•'- p!I d♦

=- a3 f dcp + a 3
2n J ~ =0.
2n

(J 0 Ans.
62
Mathematical PliYsics
. Vect,

A(c)
dV The
. d Volume
f' Integratio n. If we consider
· a closed surface in space enclosing a volume v th
JJJ is e med as the volum e m
. tegratlon.
. . co-ordinates this is written as
In cartesian , en

J) ) , Adxdydz= J J J (Axi+Ayj+Azk)dxdydz X y '

ofth
face
=if) f,Axdxdydz+jf f JAydxdydz+kf ff A,dxdydz · . . ( l ·110) nom
draw
Ex. Evaluate fff F d:, w- here F = 2xzi. - xxJ.Y+ zy2 k and V is thex region
Y z
bounded by the surface flux
V
2
x = 6 , Y = 6, z = x and z = 4. z and
th Here, th e 1~mits of x are from O to the line in which the plane z = x 2 meets
1 z=4
e Pane z = 4, i.e., the limits of x are from Oto 2, the limits of y are from O y=6
to 6 and the limits of z are from x2 to 4. surf
Thus, we have y flux
2 6 4
J J J FdV=f J Ji(2xzi-xj+y2k)dxdydz
x Y z
0 0 X

= .f J 2xz dx dy dz - j J J x dx dy dz + k J J J, y2 dx dy dz
2 6 4 Fig. 1·50
J J2 6 4 2 6 4
1 2 2
0 0 x OOx 00£

and

={32- f) ·
3
6-j(8-4) · 6+k(s-¾)- 72~ 128i-24j+384k. or

wh
§ 1-23. Gauss's Divergence Theorem. Gauss's divergence, theorem enables us to transform a
volume integral into surface integral. It states that the volume integral of the divergence of a vector field
rou
V taken over any volume v is equal to the surface integral of V taken over the closed surface surrounding
su r
the volume v, i.e.,
fJJ div V du= JJJ V · V du= JJ V · dS = JJ V · n dS . .. (1 ·111) int
u u s s in

Proof. f[f div V dv = fff (";, +~+ ":,')dxdy dz, to

where du= dx dy dz (small volume element at the centre of which the value of vector is V)

or fff div V dv = fff a;, dx dy dz+ fff ~ dx dy dz+ Hf a:,, dx dy dz

U U V V

The first term on right is integrated first with respect to x, i.e. , along a strip of cross-section dy dz

extending from p 1 to p 2. Wet thus have

fff ~~~ dx dy dz= Jf [Vx(P Vx(P dy dz
2) - 1)]
... (1 -112)
= ff [Vx (x2, y, z) - Vx (xi, y, z)] dy dz
 Vector Analysis 63

y
This integration is now considered in Fig. 1-51.
The tlux diverging from small volume element du is the surface integral P1 P2
of the normal components through the surfaces bounding the volume. For the
~
face P2 the positive direction of x-component of V and the outward drawn dS 1 dS 2
normal are in the same sense and the flux is positive while for face P 1 outward
X
drawn normal and x-components of V, i.e., Vx are in opposite directions, so
flux is negative.
Vx dz dy at P1 = -Vx dSx z
and Vx <lz dy at P 2 = Vx dSx· Fig. 1-51

So at P 1 , dz dy = -dSx and at P2, dz dy = dSx. y

The incoming and outgoing fluxes across each of the common
dv
surfaces in Fig. (1-52) cancell each other leaving only the contribution of
, I
flux across the boundary of the surface enclosing the volume v. - -r- --. -I
I
Hence from equation (1·112), we have X

HJ aa; dx dy dz= ff V:r dSx.

V S
In the same manner, we have Fig. 1-52
Hf ~dxdydz=Jf VydSy
V y S

and ffJ aavz dx dy dz =ff V, dSZ '

V Z S

f[f (~'; + ~ + aa~}1x dy dz~ f[f div v du ~ff (V, dS, + v, ds, + v, dSJ
=ffV - dS
s
or JfJdiv V du= fJf V · n dS, ... (1-113)
V s
where n is outward drawn normal. This is Gauss' theorem of divergence.
§ 1-24. Stoke's Theorem. This theorem states that the tangential line integral of a vector function
round any closed curve is equal to the normal swface integral of the curl of that function over any enclosed
surface which has the curve as its bounding edge.
Proof. The tangential line integral of a vector V around an (x, y + t:..y) (x, t:..x, y + t:..y)
infinitesimal rectangle of sides tl.X and t:.y lying in the xy-plane as shown y o ,e-'--------,"c
in Fig. l ·53 is f V. di around this rectangle. The various contributions
to .dus integr~ are as follows : ,i,
A :. B
along AB: v( tl.X,. (x. y) (x, + t:..x, y)

along BC:
➔
(v,+ ~>+Y, '------ --~x
Fig. 1-53

along CTJ: - Vx + avx

ay t:.y & , J
J (
➔
along DA : -Vy !J.y.
 64
Mathematical Pl1 .
Ystcs

So ~ V · di= (a
-! avxJ
~ - ~ Ax .1y = (V x V) z Lix .1y = curlz V dSxy,
ABCD X uy ·
where dSxy is the area of the rectangle.
Now we consider a closed curve in the xy-plane as shown in Fig.
y
1-54. We divide the space inside C by a network of lines forming a network
of infinitesimal rectangles . If we take the sum of the line integrals around
the various meshes , we obtain

.r, ~ V · dl, = .r, curlz V dS).y· .. . (l-114)

r=I r=I

Now the contributions to the line integrals of adjoining meshes X

neutralize each other because they are traversed in opposite directions; the Fig. l ·54
only contributions which are not neutralized are those on the boundary C of the surface. So

f V · di = f V · di.
00

1: . .. (1 · l 15)
r= I c
The sum on the right of ( 1-96) reduces to the following integrals:
1: (V x V)z dSxy = JI (V x V)z dSxy
s
So pV · =ff (V x V)z dSxy
dl, ... (1 ·11 6)
l' ,\'

Thus the line integral of a vector field V about a contour C of a plane surface Sis equal to the surface
integral of the normal component of the curl of V on the surface throughout S.
In fig. 1-55, we see that
z
f V-dl= f V-dl+f V - dl+f V-dl . . . (1-117)
ABC OBC OCA OAB
since the contributions along the lines OA, OB and OC cancel each other.
But using equation (l · 116) for each integral of (1 · 117), we may write:
f V • dl= II (V x V)xdydz+ ff (V xV)ldzdx+ fJ (V xV)zdxdy B
ABC OBC 4 OCA OAB
... (1·118)
But vector area dS is written as
dS = i dSx + j dSy + k dS2 = i dy dz + j dx dz + k dx dy X
Fig . 1·55

So equation (1 -118) gives pV · di= ff (V x V) · dS. . .. (l-119) z

· ABC ABC
C
In Fig. 1·56 we may regard the surface Sas being made up of an infinite
number of elementary triangular surfaces.
So L pV . di= L ff (V x V). dS from (1·119)
Ws
J..------Y
r r

h of the line integrals of the elementary triangles reduces to

. ~ut t e ~um bout a periphery C of closed surface S. So we have x
a lme rntegrat10n a
1: f
V-d/= V - di. p . . . (1 ·120)
Fig. 1-56 I
r r c
 Vector Analysis 65

In limiting situation equation ( l -119) may be written as

t
r
f V · di= f f ('v x V) · dS
r
.. . (1-121)

By the help of equations (1·120) and (1·121), we may write

f V · di= ff (V x ~) · dS. . .. (l· 122)

C

This proves Stoke's theorem. The conNerse theorem is necessarily true, i.e., if the line integral of V
round a closed curve is equal to the surface integral of A over any open surface bounded by the curve, then
A= curl V. · ··
When curl V = 0, then from equation ( l · 122), closed path line integral of V vanishes or V may be
wtitten as the gradi_e,nt of some scalar function . · "' . ., ~.. .,
§ 1-25. Green's Theorem. This theorem was first devised by Green, and commonly called after
him. This enables us to express an integral taken over the surfaces of a number of bodies as an integral taken
i) through the space between them.
Thus if u, v, ware continuous functions of the cartesian co-ordinates x, y, z, then

L ff (lu +mu+ nw) ds ~- ffJ (:: + t t;

+ )dxdy dz. . .. (1 · 123)

Here L denotes that the surface integrals are summed over any number of closed surfaces and
S) l, m, n are the direction cosines of the normal drawn in every case from the element ds into the space between
the surfaces. The volume integral is taken throughout the space between the surfaces.
Proof. In Fig. 1-57, we take any small prism with its axis parallel
to that of x, and of cross-section dy d;;. Let this prism meet the surfaces at ~ ~_
P, Q, R, S, T, U, V, W, ... cuttmg off areas dsp, dsQ, dsn, . . . . P! ~ ~ !
In equation ( 1-123), the first term on right hand side is a R s T u V W

. ffJ ~; dx dy dz. Fig. 1·57

The contribution of the prism considered above to this term is dy dz J (du/'iJx) dx, where the integral
is taken over those parts of the prism which are between the surfaces.
du JQ du
Thus J -::;-
uX
dx = -::;- dx + JS.R -::;-
p uX
du
ux
dx + ... =-Up+ UQ - UR + Us ..•
where up, uQ, uR . . . are the values of u at P, Q, R, . . . . Now the projection of each of the areas
dsp, ds0 , .. . on the plane of yz is dy dz. So we have
dy dz= lp dsp =-IQ dsQ = LR dsR = ... ,
where lp, IQ, LR, ... are the values of l at P, Q, R, ...

So dy dz f dudx dx = dy dz(-up + uQ - uR + ... ) = -lp up dsp- lQ uQ dsQ - lR uRdsR .. .

Similarly, we may take for all prims and adding the equations of all prims along x-axis, we have

fff: dxdydz=-tf luds.

Similarly, JfJ ~: dx dy dz may be calculated by integrating on prism along y-axis and we may have
fJJ :~ dx dy dz = - L f mu ds.
66
Mathematical Physics
.

Similarly, we have fff ~;f dx dy dz = - L nw ds .

So we have HJ (:: + :~ + t; )dx dy dz =-d (/u +mu+ nw) ds.

which proves Green 's Theorem .
Another form of Green's Theorem. In equation (1 ·123), if we substitute

u=-+>~ v=-+>~ and w=-+>~

'I' ax ' 'I' ay , 'I' az ,

where <I> and \\f are any functions of x, y and z, then

lu +mu +nw=$[l~ +mt+n~ J

If we have a normal k, direction-cosines of which are/, m and n along x, y and ·z, then we have

~-~
az - ak cos (k x) , ~
dy -- ~
iJk cos (k, y) and ~-~
az - ak cos (k, z) ,

where (k , x) , (k, y) and (k, z) are the angles which k makes with x, y and z-directions respectively.

Thus we have (Lu+ mv + nw) = <!> %'f- [L cos (k, x) + m cos (k, y) + n cos (k, z)] .
But from definition of directions-cosines, we have
L= cos (k, x) , m= cos (k, y) , n= cos (k, z).

and hence (Lu+ mv + nw) = <!> ~ [cos 2 (k, x) + cos 2 (k, y) + cos 2 (k, z)] = <P %'f-,
since cos2 (k, x) + cos2 (k, y) + cos 2 (k , z) = 1, from the direction cosine law. Now according to equation,
au+ av+ aw= a<1> ~+~~+ a<1> ~+<!>(~+~+~]
ax ay az ax ax ay ay az az ax2 ay2 az2 '

we have fff (<!> V \\f + i)q>ax ~

2
ax + ~
ay ~
ay + ~
az az ~J dx dy dz= -I.<!> ff ~
ak ds. . .. (1 -124)

This equation is true for all values of<!> and 'V , so that if they are interchanged and equation remains
true, then

ff (\\f y7 2 <!> + ~ ~ + ~ QP. + ~ QP.J dx dy dz= -I. ff 'V aq>ak ds . .. (1 •125)

ax ax ay ay az az
On subtracting equation ( 1-125) from equation ( 1-124), we have

ffJ ($ v¾,-w V $) dx dydz =-L ff Hf-w::]ru

2

or fff (<!> v'~ - 'V v' <!>) dx dy dz= ff (q> grad grad<!>) · dS
2
\j/ - \j/ ... (1 -126)

[ by the defi nition of gradient,~= grad \JI, :: =grad~]

 L
67
Vector Analysis

Cor. I. To derive Green's Theorem from Gauss' Divergence Theorem.

According to Gauss's divergence theorem, we have
Hf (div V) = f V · dS
u

If we substitute in this equation V = <I> V\j/, then

v -v_-avx
ax- +avy
ay- +av,_
az- -ax-a (<I> ~)
ax + ay 1-("'ay~) +a1-("'
'f' ~)
z 'f'a z

= [(~J
ax (~J
ay2 (~J]
2az2 + + q> + ~~
ax ax + ~~+~~
ay ay c)z c)z
Putting this value in Gauss' theorem, we have
HJ (<I> v2w + V<!> • V\j/) dv = H(<I> V'I') • dS
s
. . . (1·127)
u

Interchanging <I> and \j/, we have

JJf ('II V2<1> + V\j/ • Vq>) dv = JJ ('II Vq>) · dS ... (1·128)
u
On subtracting equation (1·128) from (1·127), we have
Hf (<I> V \j/ - \j/V q>) du= H(<I> V\j/ - \j/ Vq>) · dS
2 2

u
which is Green's theorem given by eqn. (1·126).
Cor. III. To derive Gauss' Divergence Theorem from Green's Theorem.
If we put <I>= 1, 'I'= \j/ in equation (1 ·126), then we have

Hf V \J' dv = JJ (grad \j/) · dS

2
or fff div (grad \J') dv = ff (grad \J') · dS
u u

If we have a vector field V = grad \j/, then this equation leads to

Hf div V du = ff V · dS = ff V · n ds .
u

which is Gauss 's theorem of divergence.

§ 1 ·26. Gauss' formula of Electrostatics from Gauss Divergence Theorem. Gauss' formula states
that Total Normal Electric Induction through a closed surface is 4n times the charge enclosed in it.
Proof. Let S be a closed surface and r denote the position vector of
any point P (x, y, z) measured from the origin 0. Let us consider small P(x, y)
element dS of the surface S · n is the positive normal (unit vector) for the
surface dS. Then normal electric induction through dS is E • dS where E is
the electric intensity at P, i.e., E = (qlr2) k, where q is the charge at the point
0 and k is the unit vector in the direction of vector r, i.e., k = -,r = .!._
rl r
So normal electric induction through dS is
0
?r · dS =? r · n dS Fig. 1-58

Total normal electric induction through the closed surface s

= fJ !L r · n dS = q ff r ·
,.3
0
,3
dS.
68
Mathematical Phy sics
.·

By Gauss ' divergence theorem, we have

Vee

ff "/ dS= f[f div(~ )du

But V -(~) = (Vr- 3) • r+ , - 3 V . r = -3,-, r .r + 3,- 3 = 0,

provided r * 0 in V, i.e., provided O is outside of V and thus outside of S.

Hence, total normal electric induction across a closed surface is zero if no charge is enclosed inside
it. whi,
If O is . 'de S , surround Oby a small circler of radius p. Let T denote the region bounded by Sand
. ~nsi
r . Then by divergence theorem,

ff n / dS = n ff °/ dS + ff n / dS = fff "v · r du =0, 3 wh,

s+r r s r r I T r
since r -:f. 0 inside T.
tra·
Thus ff n r'/ dS = - ff· nr. r dS 3
s r
Now on r , r = p, n = - -pr .

n -r
(__pr). r = r-r_ i,_ 1
So p3 - p4 - - p4 - p2 .
r3

So ff nr3. r dS =- ff nr3. r dS =ff dSp2 =J_p2 J.ff dS =41tp2

p2
=41t.
s r r r
Hence total normal electric induction across the closed surface S enclosing the charge q is

qJff n ·3 r
--~=~=~ti~c~~~~~ . . . (1·129)
s r
§ 1-27. Green's Theorem in the plane.
Statement. If S is a closed region of the xy-plane bounded by a closed curve C and if Mand N are eq
continuous functions of x and y having continuous derivatives, then

f (M dx + Ndy) = ff (~~ - tMy )dx dy,

C S
.. . (1 ·130)
w
where C is traversed in the positive ( counterclockwise) direction.
Proof. Stoke's theorem states that
f A · dr = ff (V x A) · n dS.
C S

If the curve C is plotted in xy-plane then n = k and A= Ax i + Ay j, for this plane.

So
f A . dr = Jf (V x A) · k dS. a
V
C S

Let Ax= Mand Ay = N, so that A= Ax~+ Ayj, then t

. V xA= -
aN i + aM
az ax - aM)
az J.+(aN oy k
\ ~cror Analysis 69

Thus (v' X A) . k = aN - aM and A · dr = M dx + N dy .

ax ay
So Stoke·s theorem states that

f (M cfr + N dy ) = ff (t: -tM)

C y
dx dy
S

which is Green ·s theorem in a plane. y

Ex. Verify Green '.s theorem in the plane for (1 , 1)

f [(xy + l) dx +x2 dy]

C
2
where C is the closed curve of the region bounded by y = x and y = x .
y = x and y = x2 intersect at (0, 0) and (1, 1). The positive direction in o~'-------.--x
traversing C is shown in the adjacent diagram. Fig. 1·59
Along y =x2, the line integral equals
(
0
(xx2 + x4 ) dx + f' x2(2x)
0
dx = r
0
(3x 3 + x 4 ) dx = 12_
20
Along y = x from (l , 1) to (0, 0) the line integral equals
0 0 0
fI
(x2 + x2) dx + f I
x2 dx = f
I
2
3x dx = -1.

. ed 1·me mtegra
Th en the requtr . I = 19 - 1 = - 1
20 20

We also have ff (aN

s
ox
- oM)
oy
dx dy = ff [j_ (x
ax
s
2
) - j_ (xy +
oy
l)] dx dy

1
Jf (x-2y)dxdy=(=OI;=l (x-2y)dydx=( (xy-y2) lx2
dx=-
20
.
S X Proved.
§ 1 ·28. Solution of Poisson's Equation using Green's Theorem. The general form of the Poisson's
*
equation is v' 2cj) 0 where cj) is a scalar function and v' 2 is Laplace's operator.
The Poisson 's equation is not homogeneous.
Let us suppose that Poisson 's equation may be written as
2
v' <l>=Jtx,y,z) . . . (1·131)
where/ is some scalar function of x, y and z.
In order to solve the equation (1 ·131) we recall Green's theorem which states V
that for any two functions of space co-ordinates u and w which are finite, continuous
and have first and second derivatives continuous, we have
fff (u V 2
w -
2
w v' u) du = ff (u v'w - w Vu) · dS . . .. (1 · 132)
V S
The existing problem may be dealt as if we are given the function ftx, y, z)
and we propose to find q>(x', y', z') for a fixed point of observation p'(x', y', z'), while
we have a variable point p(x, y, z) over which integrations are to be performed.
Let us consider a volume Vas depicted in the adjoining figure . It is bounded
by the outer surface S2 and the inner surface S1, a spherical cavity of radius S0 about
the fixed point p'.
Fig. 1·60
 Mathematical Physics

70
If we specify the function " by
l _!
"=Ir - r'I - / ,
•· o' then
➔ d pp'= L with respect to an arbitrary ongm
➔ ,_
r'
where Op = r, Op - an J

u= ~=~=-=x~Y?+
=~d- ~f~f~+~~~-~z~'f~] ]
so that v'u =v' (f)= v'[ ✓1(x-x')' + ~ y')' + (z-z')'J =o (y

. ti,es Laplace equation.

It follows that u as defined above sat1s
Now the equation ( l · 132) reduces to ( 1]]
fJJ v2w dv = JJ
I s
[Vt - w VI . dS .. . (1 ·133)
II

If we interpret was cl> and replace V w by

2 fin accordance with (1·131), we have
V ( l]]
f f{x, y, z) dv = J[¥- <I> v
1 . dS
... {1 ·134)
11
I s
The right hand side of the equation ( l · 134) consists
• • teg rations ' one over the surface S1 and
of two m
the other over the surface S2; we thus have

R.H.S. of eqn. (I 134) ![¥ -♦ f)l

= V( dS + !, [¥- + f)l V( dS

0
Vq> . dS 1 approac hes to 4> I · dS 1 as s tends to zero • the minus sign coming from the fact that dS
-a;
r p'
is inward with respect to the cavity. Hence

~ 41t S5
J -V<!>- S-. dS,
'- ➔ -
iJr
1
, So
➔
0 as S ----,
O
-----'- 0 •
s, p

The second integral on the R.H.S. of eqn. (1 ·134), when taken over S 1, reduces to the following form
in the limit S0 ➔ 0;

f v(½) ·dS ➔ -<l>I ,(- ~]

s,
cl>
So
(-47t S5) =-41t
P
<I>
p'
= -4n <l>(x', y', z') .

Hence as S0 ➔ 0, the equation ( l ·134) reduces to

[ fdv=-4n ♦(x',y',i)+ f,[¥-+v(+)]ds .. (1135)

Here the remaining integral over S2 on the right has a simple meaning. It is a solution of Laplace's
2
equation in the form V cj>(x', y', z') = 0; for the only quantities which depend on the fixed co-ordinates are
1// and V(l//), and these clearly satisfy it. Hence if this whole integral were subtracted from</), the remaio<ler
would still satisfy the equation (1·131).
We now allow v to become infinite and impose on <!> the boundary condition that at infinity it vanis~es
as 1/rdoes. Then V4>/l and cl> V(l/r) are both of order l/r3 at infinity and after the surface integration which
amounts to multiplication by ?, the result will still be of the order of 1/r and hence vanishes.
Vector Analysis 71

Of interest, therefore, is chiefly the particular solution which re ma ins whe n the integral over S2 in
equation (l · 135) is omitted, i.e. ,

J +f (x , y, z) dv = - 4mj> (x' , y', z') .

It is usually known as the solution of Poisson 's equation.

Thus "' ( ,
'I' x , y ,, z') = _ _4l7t J f I(x, y, 'Iz) dx dy dz. ... (1 ·136)
r- r
§ 1·29. Classification of Vector fields. We have seen that if curl V = 0, then V = grad s or zero-curl
field and if div V = 0, then V = curl A or V is a solenoidal field . In practice there are the foll owing fo ur
classes of vector fields:
(1) curl V =0 and div V =0. According to first condition the field is lamellar
and irrotational and according to second condition the field is solenoidal or
incompressible.
So from first condition V = grad S and from second condition
div grad S = 0 or V 2S = 0 which is Laplace equation. The example of this type of No div, No curl
field is irrotational r,10tion of incompressible fluid as shown in fig. l ·61. Fig . 1-61
(2) curl V = 0 but div V * 0. Field is irrotational but not solenoidal.
From first condition, we have
Applying second condition, we have
V = grad S.

div grad S -:t 0 or V2S -:t 0.

I•
This is Poisson's equation.
Example of this type of field is irrotational motion of compressible fluid as
shown in Fig. l ·62.
(3) curl V * 0 and div V = 0. The field is rotational and solenoidal.
l
Fig. l ·62
~
~

II\
Second condition states that V = curl A ;
then curl curl A *0 according to first condition.
So grad div A - V2A -:t 0.
If A is solenoidal, then div A= 0 or we have V 2A * 0 as determining equation
for A. The example of this type of vector field is rotational motion of
incompressible liquid as shown in Fig. 1-63. \ I

(4) curlV-:t0anddivV-:t0.
Fig. l ·63
This is the most general type of a vector field as rotational motion of
compressible fluid. This field V can be expressed as sum of two fields; the first is
lamellar vector which has no curl but can have divergence and the second is a
solenoidal vector which has no divergence but can have curl. So
V = grad S + curl A, .
i.e., a vector function can be expressed as the sum of a lamellar vector and a
solenoidal vector.

1 64
This is Poisson 's equation and determines S. Fig. ·
curl V = -V 2A * 0 if A is solenoidal.
This equation determines A. This decomposition of a general vector field into two fields one being
lamellar and other solenoldal is called Helmholtz's Theorem. The example of th is type of field is rotational
motion of compressible fluid as shown in Fig. l ·64.
 Mathematical P/zy, .
72 sics

In a general way Helmholtz 's theorem states that a vector V satisfying the equations
V • V = s( r) and Vx V = c( r) . . . ( l ·13 7)
with both the source (charge) and circulation densitites s and c respectively vanishing at infinity may be
written as the sum of two parts one of 1,1,·hich is irrotational and other solenoidal.
for the rigorous proof of this theorem we prove another theorem known as uniqueness theorem which
states that a vector is uniquely specified by giving its divergence and curl within a region and its nonnat c
component over the boundary. For proving this theorem let us assume that the vector V1 has divergence and
curl given by equations (1 ·137). Let us also assume that the normal component Vi on the boundary is given. a
Suppose there exists some another vector V 2 which satisfy equation ( l · 137) and has the same normal rt
component over the boundary.
Let us define then W=V 1 -Vz.
Obviously V·W=V•V 1 -V•V2=0 .. . (1 ·138)
and V X w = V X VI - V X V2 = 0. . . . (l ·139)
Equation ( l · 139) shows that W can be written as the gradient of a scalar field, s
W=-Vcj>. . . . (l · 140)
which when substituted in equation (l · 138), gives
V · Vcj> = 0 (Laplace's equation) . .. (1 ·141) d
. Moreover, Wn =Vin - V2n =0. . . . (1 ·142)
Let us consider now a vector U = <I> Vq, which gives
div U = V · (<I> Vcj>) = cj>(V. V<j)) + (V<I>) . (V<j>).
Surface integral of this vector U when reduced to volume integral using Gauss divergence theorem a
becomes

ff ¢ V
S
ff f [{cj>(V · V
<I> · dS =
V
<I>)} + V<j> . V <I>] dv

or fff (V</>) · (Vcj>) dv =ff (</> Vq,) dS using eqn. (l · 141)

V s
=- ff <l>WndS=0 using eqn. (1 ·142)
f<
Thus fff W · W dv = 0.
V
d

which gives W=V _V _

. . 1
0 2- or V1 = V2, everywhere.
Thus VI ts unique, proving the theorem and h th . . .
(l · 137) (and boundary condition) Helmh It , th en~e I e vector. V is umquely defined by the equatwn
· 0 z eorem 1s c early satisfied if
,
. . . V = -Vq, + V x A, . 43) a
where -Vq, is motational and V x A is solenoidal. Let us ·u . .. . . .. (1 1
From the functions s(r) and c(r) of eq f (l J Stlfy this equation to prove the theorem.
such that ua ions · 13 7 ), we construct the scalar (j>(r) and vector A(r)

q>(r) = _l
4n
f s(r')l dv' . .. (1 -144)

A(r) = ...!_ f c(r') dv'

4n l . . . (1 -145)

h
 73
Vector Analysis

where r , r' and l are shown in Fig. 1·60 of § 1·28 with S1(x' , y', z') the source point and Sz(x, y, z) the
fie ld poi nt. 01' course, s and c must vani .:;h sufficiently rapidly at large distances so that the5e jntegrals
exist. From equation (1 ·143), we have
. .. (1 •146)
V · V = - V -V4>
... ( l · 147)
and V xV=VxV xA .
. . . (1-l 46a)
Thus if we can show that - V. V4>(r) = s(r)
. . . ( l · 146b)
and V x V x A(r) = c(r) .
then V gi ven by (l · 143) will have proper divergence and curl and it will be justified .
From r,quation (1 ·144) and (l-146),
V · V = -V · V4> = - l V. VJ s(r') du' . . . ( l·l46c)
4n l

=- i,. f s(r') v (t) dv', 2 . . . (l-146d)

since V 2 operates on co-ordinates, x, y, z and so commutes with integration with respect to x', y', z'.
... (1 ·147)
In § l ·26, we have shown that

depe nd ing on whether or not the integration included the origin r = 0.

But V(;)=-V 2
(~)

and hence equation (1•147) gives f V-(~)du={;'"

2

This result was conveniently expressed as

2 . .. (l· 148)
V -(~) = -41t 6(,) ,

when Dirac delta function o(r) is defined as

O(r) = 0, n~ 0 ... (l· 149)
ff(r) o(r) du = f(O), . .. (l· 150)
for /(r) a well beha ved function and the volume integration including the origin. This function will be
discussed in details in Chapter XVI.
Generalizing equation ( 1•148) for the system under consideration, we have

(!J/ 1
v 2 =V 2 ( --) =-47t o(r - r')
Ir - r'I '
. . . (1·151)

o(r - r') = 0, r :t: r' . ... (1·152)

where
and f /(r) o(r - r') dr = J(r') . . . ( l · 153)

is clear that v2(+)= V'2(+}

o(r - r') = o(r' - r) . . . (1 ·154)

Thus we may have v V= ~!J s(r')V'

2
(f)ar'
74
Mathematica[ Pl1 .
Ysics
Ve,
= :! f s(r')(-4n) 8(r' - r) dr' =s(r) (using 1·153)

which proves eqn. ( l · 146a). To complete the proof of Helmholtz theorem it requires to prove equar or
(1·146b). 10n
wh
From equation (1·146), V x V = VV. A - V2A,
· · · (1 -155)
where the first term vv. A= - 1-vv. c(r') du'
4n
f I Wht
or
41t \7\7 · A = f c(r') · \7 fJdu' .
2
( . .. (I . I56)
ins

and
On substituting V2(11L) = V'2(11L) and integrating each component of (1·156) by parts, we get

f
41t\7\7 ·A= c(r') • \7' f
d~ •( Jdu'

= f \7' · [ c(r') a~ (fJ] f du' - [\7' · [c (r')l a:,(f) du' ... (1·157)

where the second integral vanishes since V' • c(r') = 0, c(r') being solenoidal). The first integral may also
be written as

f \7' · [ c(r') a!, (fJ] du'= fs c(r') a:, (fJdS. . .. (1 ·158)

(using Gauss' theorem)

which vanishes for a sufficiently large surface S because c vanishes faster than 1/r for large r.
Thus equation (1·155) reduces to
whi
2
\7 x V = -\7 A = - 4~ f c(r') \7 2
•( ~ dv'. . .. (1 -159)
and
which is exactly like eqn. (l · 146d) except that the scalar s(r') is replaced by vector c(r'). Thus in the
manner identical to that used for equation (1·146d), equation (1 ·159) reduces to V x V = c(r) which is · whe
equation (l · 146b) and this completes the proof of Helmholtz's theorem.

SOLVED EXAMPLES I
Ex. 1. Show that each of the three equations
F=V(j> . .. (i)
VxF=O ... (ii)

f F · dr = 0 .. . (iii)
Let
implies t/ze other two. . .
It has been shown in equation (l ·55) and article § l · 16 that the given equation (i) implies equations
(ii) and (iii).
Let us start with equation (iii) which means that the line integral along the closed path is zero or th e
line integral is independent of the path and depends only on the end points, i.e.,
B
f F · dr =
A
<1>8 - <1>A
... (iv)

For the points A and B separated by the length dr this equation reduces to
 HV · ,L.~ = HV · i dy clz + JJ V · j dx dz + JJ V · i c/z dy
S AUCH /lfGC H ill/!

+ JJ V · j dx dz+ HV · k dx dz+ JJ V · k dx dy = 3
El/VA C/)//G Ali ff.

Hence HJ div V · dv = HV · dS,

u s
which proves the divergence theorem .
Ex. 20. Express the position vector V = xi + y j + zk in spherical coordinates.
Using equations ( l ·89), we have
r = r sin 0 cos <I> i + r sin 0 sin <I> j + r cos 0 k
= r (sin 0 cos <I> i + sin 0 sin <I> k + cos 0 k)
= r i,, (using eqn. 1-95)
Ex. 21. Express the vector V = 2xi - zj + yk in (a) cylindrical coordinates and (h) spherical
coordinates.
(a) Using equation (1 ·74), we get
V = 2r cos 0 i - z j + r sin 0 k
Substituting the values of unit vectors i , j and k from equation ( l ·93a), this equation becomes
r = 2r COS 0(COS 0 i, - sin 0 ie) - Z (sin 0 i, + COS 0 ie) + r sin 0 ik
= (2r COS2 0 - Z sin 0) i, - (2r COS 0 sin 0 t Z COS 0) ie + r sin 0 ik
which is the expression for the given vector in cylindrical coordinates.
(b) Using equations (1·89) and (1·95b), the given vector may be expressed in the spherical coordinates
as fo\\ows:
V = 2r sin 0 cos (\>(i, sin 0 cos <I>+ i8 cos 0 cos <l> - ¼sin <l>)
- r cos 0(i, sin 0 sin <l> + ict, cos 0 sin 0 + ict, cos <l>)
+ r sin 0 sin <l>(i, cos 0 - ie sin 0)
2
or V = 2r sin2 0 cos2 <I> i, + (2r sine cos 0 cos <I> - r sin <I>) i0
- (2r sin 0 sin <I> cos <l> + r cos 0 cos <I>) i~
Ex. 22. Find the time derivatives of unit vectors i,, ie and i, of cylindrical coordinate system and
obtain the expressions for position vector, velocity and acceleration.
From equation (l ·93) in cylindrical coordinate system, we have
di, . d0 . d0 •
dt = - sm 0 dt 1 + cos 0 dt J

de . . .) de . .
= dt (-sm 0 1 + cos 0 J = dt 10 = 0 le . .. (1)

-die = -cos e -d0 .l - . d0 . d0 ( e . . e .)

sm 0 -d J = - -d cos l + sm j
dt dt t t
= -8 i,;
di,= O
and dt
The position vector r may be written as
r=xi+yj+zk
= r (i cos 0 +j sin 0) + zk
=ri,+zi,
Vector Analysis 93

In case motion is steady, aq

at =0 or ~=0
at .
Then 13(t) = 13 (a constant) and hence equation (l · 181) reduces to

V + !!_ + .!. q2 =~(constant) . . . (1 ·183)

p 2
which is known as Bernoulli 's equation which states that per unit mass of fluid, the sum of kinetic energy,
potential energy V and the pressure pip have a constant value for all points of the fluid.
The general equation ( l · 179) may also be written as

2q xiii'= v(v+ J! +½ q'} ... (l · 184)

since l vp = v f dp
p p
Multiplying equation (1·184) sca!arly by q, we get

2q qxro'=q • v(v+ Jt +½q') or v(v+ f b+½q')=o .. . (I 185)

[since q · (q x ~ = 0.]
dp l
which gives V+ f p+ 2 q =c, 2 .
(for steady motion) . .. (1·186)
where the constant c depends upon a particular stream line chosen. This equation is known as Bernoulli s
Theorem.
Let us consider equation ( 1-183) (Bernoulli's equation) at the free surface Y ',
and at the horizontal orifice in the base of a vessel filled upto the height h with ' ',
the fluid of density p as shown in the diagram. The vessel is constantly full of
the fluid upto height h. The cross-section 'a' of the orifice is much smaller in
comparison to the area 'A' of the free surface. Suppose ' V' and ' v'are the
velocities of the free surface and the orifice.
Measuring z-downward, we have V=-gz

and hence eqn. (1·183), becomes -p + -l q-1 - gz = ..,A . . . . (1·187)

p 2
I \'
If Po is the atmospheric pressure, then equation ( l · 187), for the free surface z
is given by
Fig. 1·69

~o+½l=~. (since z =0 at free surface)

or ;+½v2=f3. . . . (1 ·188)

Equation ( 1-187) at the orifice becomes

; + ½v
2
- gh = [3 , (since z = h at orifice) . . . (1·189)

1 Po l yi
Thus -Po + - V 2 - gh = - + - or ... ( l · 190)
p 2 p 2
Condition of continuity requires that

AV=av or .. . (1·191)
94

Substituting thi s value in equation ( l · 190), we get

Mathematical Pl l
lYsics

=2gh [neglecting square and hi gher powers of a/A]

or v = ✓(2gh) · · · 0 ·192)
or voe%,
i.e., velocity of flow of the fluid through orifice is proportional to square root of the height of free surface
above the orifice. This is Torricelli's Theorem.
Viscosity. A perfect fluid is assumed to have no property of the nature of friction , i.e., incapable of
exerting shearing stress whether at rest or in motion and therefore pressure at any point in such a fl uid is
the same in all the directions. In practice no fluid is perfect and it offers resistance to shearing stress. This
characteristic of the fluids is known as viscosity. Taking this property into account the equations of motion
in hydrodynamics are considerably modified because in a viscous fluid both the tangential and normal forc es
exist.
Let us assume that the moving fluid is divided into parallel layers which flow past one another with
various velocities which decrease as we go downward such that velocity of the layer in contact with the
horizontal bounding surface is zero. Let us consider a layer at height h from the horizontal boundary and
moving with velocity V. Then the uniform velocity gradient is Vlh and the horizontal force required to
maintain the motion of this layer is proportional to Vlh for the unit area of the layer. Thus this horizontal
force for the unit area of the moving layer under consideration is
V
ri h,
when 11 is called the coefficient of viscosity. If the velocity gradient is not uniform then this horizontal
force for the unit area of the layer is
du
ll dz '
where z-axis is taken ..L to the layers and u is the velocity at any point at height z from horizontal boundary
surface. Thus the viscous force on the area S of the moving layer is
du
F=17S dz. . . . (1 ·193)

For calculating the force of viscosity per unit volume acting on the fluid we consider an infinitesimal
element of the fluid with volume ◊V enclosed by surface ◊S such that point A is its centroid and the point
Bis on the surface as shown in Fig. 1-70. The fluid velocities at A and B are q and q + dq .
oq oq oq
Let AB= r, then dq=x-+y-+z-
ox oy oz
q
..y + zk)(o.
= (XI+ o . ok]
. OX I+ Oyj + OZ
J Q
B

= (r · v') q . . . ( 1•194)
which is relative velocity of the fluid particles B and A.
Obviously ~he equation {(r · v') q} • r = c (constant) .. . (1·195)
Fig. 1-70
being quadratic in the components of r , represents the equation of the surface under
consideration.
At a parallel surface the infinitesimal distance dr from B, the similar surface equation is satisfied, i.e.,
[{ ( r + dr) · v' }q] • (r + dr) = c
Vector Analysis 95
{(r-V)q} · r+{(dr-V)q} · r+(r - V)q} •dr= c,
or d · · • (I · 196)
. Vo erator operates on q an not on the constant vector r) , wh ich on usi ng cqn. (1 ·61) , becomes
(since P dr · {r x (V x q) + (r • V)q} + {(r • V)q} . cir= O
dr · {r X (V x q) + 2( r . V)q} = 0.
or
f(r) = 1,1 { r X (V X q)} + (r · V)q
Thus the function . . ... (l -197)
· e in the direct10n of q + dq.
is normal to c.I r, ' · .,
From equation ( 1-197), we have
(r · V)q = f(r) - ½ { (V x q) x r}
which when substituted in ( 1-194) gives,
dq =f(r) + 1/2 {(\7 x q) x r} ... (1 -198)
Thus the relative velocity of B with respect to A consists of two parts:
(i) angular velocity of rotation l/2 (V x q).
(ii) function f(r) which is known as pure strain.
Angular velocity of rotation of the whole element under consideration causes no relative motion. It
cannot produce a force of frictional nature and therefore the viscous forces are attributed to the pure strain
f(r) .
The stress force (viscous force for unit area) due to this strain is proportional to strain f(r). Thus the
force exerted on an element 8S is 2T) f(r) 8S, where 'f1 is the coefficient of viscosity. Hence the force of
viscosity per unit volume is

T = Lim s:.lv
IW ➔ Ou
J 2rt f(r) dS
S

=
6V ➔ O
J rt 2{ (r · V)q + -21 r x (V x q)} dS
Lim s:.lv
u S
(using eqn. l · 197)

= Lim 8 V 2T) J(r · V)q dS + Lim sP- J r x (V x q) dS.

1
. .. (1-197)
6V ➔ O 6V ➔ O
uV
But we have

Lim -f- fs r x A dS = curl A

6V ➔ O uV
and Lim
6V ➔ O U
}V JS r • A dS = div A ,
where r is the unit vector normal to the surface S. Thus equation ( 1-199) reduces to
T = 2T) V 2q + rt \7x (V x q) = 2rt \72q + rt \7(v' · q) - 'f1 V2 q
... (1 ·200)
(using eqn. 1-58)
Hence the force due to viscosity acting on the unit mass of the fluid , T/p (since T is the viscosity
force acting on the unit volume or
fluid), is given by

T = .!l. v72q + .!l. \7(V . q) = u V 2q + v \7(\7 · q) , .. . ( 1-201)

p p p
where V == 'flip, is defined as coefficient of kinematic viscosity.
. Thus in viscous fluids the force given by (1-201) also acts on unit mass in addition of force F, ~sed
m eqn. (1-177), acting on the unit mass and hence the Euler 's equation of motion ( 1· 177) must b~ modified
for viscous fluids by addition of force given by (1 -201) to the force F (acting on unit mass) as follows :
dq T 1 7 \i'p ... (1 -202)
- = F + - - - v' p = F + v v7-q + V (v' · q) - ·
dt p p P
. This equation is known as Navier-Stokes equation of motion for a viscous fluid . For incompressible
fluid this equation reduces to
96 Mathematical Pl1 .
Ysics

dq = F - Vp + v V2q
dt P · · · 0 ·203)
oq Vp 2
which can also be written as - + (q · V)q = F - - + v V q. · · · (i -203a)
a, P
Cartesian equivalent of which is given by the following equations.
Du l ap Dv l ap 2 Dw I dp
- = F ---+vV w
2
-=F.---+uV 2u, -=Fy---;-+vV v,
Dr ·' p ox Dt p uy Dt z p dZ

. .. (1·203b)
Navier-Stokes equation (1-203) for incompressible fluid (i.e ., V · q = 0) reduces to the follow ing forms
when the motion is steady (i.e ., for dqldt = 0):
Vp
F--+uV 2q=O or F- Vp + .!l. V 2q = 0. .. . (1 -204)
p p p
When the fluid is free from external force like gravity (i.e., F = 0), this equation reduces to
Vp = rJ V2q. . .. (1 -205)
Let us assume now that the solution of V • q = 0 is
q = Q1 + 42 ... (1·206)
where q 1 = Vq> (potential flow) and q 2 is not derivable from a potential.
Then equation ( l ·205) becomes
v'p = rJ V2(grad <I>)+ rJ V 2q 2
2
=11 V (v'(j)) , . .. (1 -207)
where it has been assumed that V2 q 2 = 0.
Integration of equation ( l ·207) gives

p =11 v' 2 + C =l1 v' 2<!> + Po

<1> ... (1·208)
where the constant c of integration has been chosen in such a way that V 2<j> = o for p =Po •
Moreover, equation (1·206) gives
2
div q = v' (j) + V · q2 = 0. . .. (1 -209)
Let us assume that q2 = u2 i, such that
2 2
V u2 =0, as V q2 =0 by assumption, or div grad u2 = O. ... (1 -210)
For a sphere which is falling through viscous fluid, u 2 depends on r only and
du r
grad ui(r) = -d . - (using eqn. (1 -46))
r r
and div grad U2(r) =-\ aa ·(r du 2)
r r
2
dr (using eqn. (1 -85))

d2u2 2 du
=--+--
dr2 r dr ·
Thus equation (1·209) for a sphere becomes
2
d u2 2 du
dr2 + ~ dr =O
which gives on integration du2 _ A
dr - ,-2 '
where A is constant of integration.
 Vector Analysis
97

A
Integrating again, we get U2 =- -+B
r
but B = 0 since u2 ➔ 0 as r ➔ 00
•

Thus solution of equation ( l ·210) is u2 =-ar where a= -A ' . .. (1·2 11)

q2=-1
a. ... (l -2 ll a)
and hence r
·203b)
. forms or div q2 = ~i;r,) x
. .. (1 -212)

1-204)
Thus equation (1 -209) becomes V2q> +a_!(.!.)=
ax r O,
. . . (1-213)

solution of which is "'=_!I:

v a,
2 ax.
... (1 ·214)

1 -205) For q2 to be solenoidal (so that q can be written as sum of irrotational and solenoidal vectors)
V2q> = O (from equation (1-209))
1·206) For <I> given by (1 -214) to satisfy this condition, suitable terms Ux and ~[a(l/r)/ax] should be added,
i.e.,
<I> = _ !!: a, + Ux + ~ a( 1/r) , .. . (1 ·215)
2 ax ax
q= t'7
v ( a a, A a( 1/r) + uX ) +-1=ul+VJ+W
---+..., ex.. • • k , . . . (1-216)
) -2 07) then 2 ax_ ax r
(using equations (1 -206), (1·211a) and (1-215)).

u = U+~ a2(l/r) - !I: a2, + !I:

)-208)
which gives ax 2 ax
2 2 r

=U+~(3;' _~)-*-~)+~, ... (1 ·217)

:t -2 09)
... (1·218)

:1-210) .. . (1·219)

. aay2(1/r) ~ a2, ~ J
( I -46)) (smce ax = ~ , ay ax = - ? etc.

If we apply the boundary conditions that at r =R (radius of sphere), u =v =w =O for all x, y and z,
the equation ( 1-21 7) gives
~ (3x2
2
U + a. +
R
__!_) =!I:(..!. _ x J
/?5 R 2 R R 3 3
'

which is true for every x. Equating the coefficients of x2 on both sides, we get
2
3~ ex. aR ... (1-220)
Rs=- 2R3 or ~=-7
while equating other terms to zero, we get
98 Mathematical Ph .
Ysics

U+ ~ - _li= _Q_ or · · · 0 ·221)

R R3 2R
Solving (1 ·220) and (l -22 1), we get
a =-3/zUR, · · · (1 -222)
Thus equation ( 1-2 I 3) becomes

3
V 2¢, = - UR -
2 axa (1)
- .
r
Substituting this value in equation ( 1-208), we get

p = Po + i311 UR ax
a (1
-;:)= Po - 2,3· -VT)Rx
,) · · . (1 ·223)

Thi s equation is known as Stokes ' relation for pressure on moving sphere in a viscous liquid.
Substituting the values of constants a and P in equations (1-217), (1 ·218) and (1·219), we get the
components of the cartesian velocity:

? )x
2
U _ u[l - 3R4r _ 4?
R 2
)- 3UR
4? [l -R 2
... (1 -224)

v =- 3UR
4?
[1 - R2)xy
? ')
... (1 ·224a)

w= - jf ~)xz
3
(1- . . . (1 ·224b)

§ 1-31. Application of vectors to heat flow in solids. Let us consider a closed surface S enclosing
the volume Vin a block of solid of density p and specific heat cr. The amount of heat M inside this volume
of the body is given by
H= ffJ (ucrp) dV , . .. (1 ·225)
V

where u = u(x, y, z, t) is the temperature of the body.

It is an experimental fact that the rate of flow of heat into the volume dV is proportional to the product
of temperature gradient and surface dS enclosing the volume. Thus rate of heat flow into the volume V is
given by
dH =
dt s
ff
K au dS
an ' ... (l ·226)

where dn is outward normal on the surface dS and K is consta~t known as thermal conductivity.
Differentiating equation (1 ·225) with respect to time and equating to equation (1 ·226), we get

fij (c,p %i) dV = II( K ~: )dS . (1221)

Obviously, du= 'vu• dn, . .. (1 ·22 8)

where du is the total derivative of temperature, and it represents the change in temperature on moving a
distance do. This equation gives
au ='vu·n
- ... (I -229)
an '
where n is unit vector normal to surface. This equation (1 ·227) becomes