Phys374, Spring 2007, Prof.

Ted Jacobson
Department of Physics, University of Maryland

Waves on a stretched string

A stretched string will vibrate when plucked. If the string is finite it oscil-
lates, while if the string is infinite it supports traveling waves. The physical
parameters defining the problem are the mass per unit length ρ and tension
τ of the string. In addition if it has finite length then the length ` will enter
the expression for the oscillation frequency.

Dimensional analysis
p
Exercise a: Show that the combination τ /ρ has dimensions of veloc-
ity, and that one cannot make a dimensionless quantity using ρ, τ , and the
wavelength λ. This allows us to infer
p that the wave speed is independent of
wavelength and is proportional to τ /ρ.

Exercise b: If the string has fixed endpoints then it can vibrate at a

particular set of normal mode frequencies. The lowest frequency must be
proportional to some combination of the available constants ρ, τ , and `.
Find this combination.

Wave equation
Here we use Newton’s law to derive a partial differential equation describing
the motion of the string. We suppose the equilibrium configuration of the
string lies along the x axis, and we let y(x, t) denote the perpendicular
displacement of the string from its equilibrium at position x and time t. We
assume that the displacement of the string is very small, in the sense that
∂y
1 (1)
∂x
which means that the slope of the string is everywhere very small compared
to one. Equivalently the angle θ between the string and the horizontal is
small.
Since different parts of the string have different motions, we need to apply
Newton’s law F = ma to each infinitesimal bit of the string separately. To

1
this end, consider the bit of string that runs from x to x + dx. Since we
assume the slope is very small, the length of this bit of string is nearly just
dx. The correction is of order (dx)2 . Neglecting this, as well as the related
stretching of the string1 , we have that to lowest order in dx the mass of this
bit of string is
m = ρdx. (2)
The acceleration of this bit in the y direction is, to zeroth order in dx, given
by the second partial derivative with respect to t,

∂2y
ay = (3)
∂t2
evaluated at x.
It remains to evaluate the y component of the force on this bit of string.
The force arises from the vector sum of the forces due to the pull of the
string on the right and on the left of the bit.2 The string exerts a force of
magnitude equal to the tension τ , and direction along the string. The y
component of the tension force on the left depends on the angle θ made by
the tangent to the string at x and is

Fyleft = −τ sin θ (4)

3
= −τ tan θ + O(θ ) (5)
" 3 #
∂y ∂y
= −τ +O . (6)
∂x ∂x

The minus sign is because the string on the left pulls downward (in the
negative y direction) if the slope is positive. Since we assume the slope is
much smaller than 1, the correction term can be neglected in what follows,
but it should be clear that our result for the force is only accurate up to a
correction of relative size equal to the square of the slope. The string force
1
The string mass is fixed. When the string is displaced from equilibrium and is therefore
lengthened its mass per unit length must decrease. To take this into account we would
have to allow for the string to have a nonuniform density, and also a nonuniform tension.
This would require us to introduce another functional freedom, describing the deviation
of the string from its equilibrium density at each x and t. The resulting system would
be more complicated to handle. I read that for the bass strings on a piano this degree of
freedom does play an important role in the nature and tone of the vibrations.
2
We ignore any force due to resistance to bending the string since, under our assumption
of small slope, it is reasonable to imagine that the bending force is negligible. Of course
one can consider a very stiff “string”, e.g. a fine metal rod, for which the bending force
would be important or even dominant. But by “string” we mean something that has
negligible resistance to bending the small amount considered here.

2
on the right is given by a similar expression, without the minus sign. Thus
to lowest order in the slope the net force in the y direction due to tension is
∂y ∂y
Fy = τ −τ . (7)
∂x x+dx ∂x x
This would be zero were it not for the slight difference in location where
the slopes are evaluated. Since Newton’s law equates this to may with m
of order dx, we only need to evaluate Fy to this order. To do so we expand
the first term in dx keeping only the first order term. That is, we apply the
relation f (x + dx) = f (x) + f 0 (x)dx + O((dx)2 ) to the first term. That is,

∂y ∂y ∂2y
= + dx + O((dx)2 ), (8)
∂x x+dx ∂x x ∂x2 x

so to O(dx) Fy becomes
∂2y
Fy = τ dx. (9)
∂x2
Now we impose Newton’s law Fy = may , with (2), (3), and (9). Note
that both Fy and m are of order dx, while ay is of order 1, so the equation is
sensible. Dividing by the common factor of dx we obtain the string equation
of motion
∂2y ∂2y
τ 2 =ρ 2. (10)
∂x ∂t
Equivalently, we can write

∂2y 2
2∂ y
− v = 0, (11)
∂t2 ∂x2
p
where v = τ /ρ. This is second order, linear partial differential equation is
a wave equation for waves that travel at speed v.

General solution to wave equation

The wave equation (11) can be re-expressed in the form

∂ ∂ ∂ ∂
  
−v +v y = 0. (12)
∂t ∂x ∂t ∂x
∂ ∂ ∂ ∂
The cross-terms cancel since partial derivatives commute, ∂t ∂x y = ∂x ∂t y.

3
 The two derivative operator factors on the left hand side of (13) also
commute (since v is a constant). Hence y(x, t) is a solution to (13) if it
satisfies
∂ ∂
 
±v y = 0. (13)
∂t ∂x
Solutions are given by

y(x, t) = f (x − vt) + g(x + vt) (14)

where f and g are arbitrary functions of their arguments. The f solution

represents a pattern moving in the positive x direction with velocity v, while
the g solution pattern moves with velocity −v. Note that the unknown di-
mensionless coefficient in the wave speed turned out to be 1.

Exercise c: The above reasoning implies that (14) satisfies the wave
equation (11). Verify this explicitly by direct substitution of (14) into (11).

It is not hard to show that (14) is in fact the general solution to the
wave equation. Since that equation is second order, the initial data at time
t = 0 (for example) that determine a solution are the two functions y(x, 0)
and ∂t y(x, 0). One can solve for f and g in terms of these two functions.

String with fixed endpoints

If the endpoints are fixed, the string motion is still governed by the wave
equation, but at the endpoints we have the restriction y(0, t) = y(`, t) = 0.
These boundary conditions can be satisfied by appropriate superpositions
of right and left moving traveling waves that cancel out to zero at the two
endpoints. We’ll come back to this later in the course.