GENERAL ARTICLE

The Brachistochrone
P C Deshmukh, Parth Rajauria, Abiya Rajans, B R Vyshakh, and Sudipta Dutta

The brachistochrone problem posed by Bernoulli and its solu-

tion highlights one of the most famous experiments in physics
which illustrates the variational principle. This pedagogical
study is designed to ignite a classroom discussion on the vari-
ational problem. We overview the Euler–Lagrange formal-
ism of the variational principle and obtain the solution to the
brachistochrone problem. We demonstrate the success of the
variational method using brachistochrone models which were
fabricated specially for this purpose.

1. Introduction

According to the first law of motion of classical mechanics, due to

Galileo–Newton, we know that the mechanical states of an object,
whether at rest or at constant velocity, are completely equivalent.
The frame of reference in which this equivalence holds, termed as
the ‘inertial frame’, is determined only by the initial conditions.
The object is then said to be in ‘equilibrium’ in the inertial frame.
Newtonian mechanics accounts for any change in equilibrium by
invoking a stimulus (referred to as ‘force’). The object on which
the force acts responds to it by undergoing an acceleration, which
is linearly proportional to the force:

F = dp 2
dt = ma = m ddt2r (in usual notations).

The proportionality m between the acceleration (response of the

object) and the stimulus (force) is the inertia (or mass) of the ob-
ject. This second-order-differential equation, named Newton’s
second law, embodies causality/determinism: the cause (force) Keywords
determines the effect (acceleration). Solution to this second or- Variational principle, brachis-
tochrone, tachystoptota, ex-
der differential equation requires two constants of integration,
tremum action, Fermat’s principle,
which come from the initial conditions of position and velocity Euler–Lagrange equation.

RESONANCE | September 2017 847

 GENERAL ARTICLE

Figure 1. Alternative
trajectories of a system
with one degree of freedom.
Newtonian mechanics
explains the selection of the
path taken by the object to
reach the final position q f
at a later time t f if it started
out from position qi at initial
time ti by natural laws de-
scribed by the cause-effect
2
relationship F = m ddt2q .
Hamilton’s variational
principle accounts for the
selection of the path as one
that makes ‘action’ (defined
below by (1), Section 3) (or equivalently, the initial position and momentum), considered
an extremum. Lagrange’s simultaneously knowable in classical mechanics, which together
equation included in this fig- describe the mechanical state of the object.
ure is discussed in Section
3.
The ‘variational principle’ offers an alternative formulation of
mechanics. It does not build on the principle of causality and
determinism contained in Newton’s second law. This method is
extensively discussed in several texts [1, 2]. There have been
several recommendations [3–5] in recent years to propose that
these methods be introduced rather early in college curriculum.
The formulation of the These recommendations are inspired by the fact that the varia-
laws of physics on the tional principle lends itself easily to be adapted in frontier re-
basis of the variational search areas in physics, apart from providing some distinctive
principle provides
transparent relationships advantages over the Newtonian methods. Specifically, the vari-
between symmetry and ational principle, often referred to as Hamilton’s variational prin-
conservation principles ciple, is best adapted to provide a ‘backward integration’ for the
which are of development of quantum theory which supersedes classical me-
fundamental importance
in understanding the chanics. Besides, the formulation of the laws of physics on the
laws of nature. basis of the variational principle also provides transparent rela-
tionships between symmetry and conservation principles which

848 RESONANCE | September 2017

 GENERAL ARTICLE

are of fundamental importance in understanding the laws of na- It nearly feels like a deep
ture [6]. conspiracy of nature that
the principle of
It nearly feels like a deep conspiracy of nature that the principle causality/determinism
of causality/determinism and the variational principle produce and the variational
principle produce results
results that are completely equivalent, even as the former makes
that are completely
no use of the variational principle, and the latter makes no use of equivalent, even as the
force. The essence of the difference between Newtonian formu- former makes no use of
lation and the alternative based on the variation principle, due to the variational principle,
Lagrange and Hamilton is illustrated in Figure 1. and the latter makes no
use of force.
The variational principle is also known as the ‘principle of ex-
tremum action’. The general mathematical framework for the
development and application of this technique is the ‘calculus
of variation’. Its beginnings can be traced to the solution pro-
vided by Isaac Newton, to the famous ‘brachistochrone prob-
lem’, which was posed by Johann (also known as Jean or John)
Bernoulli in 1696. We hasten to add, nevertheless, that the formu-
lation of the principle has a rich and intense history which dates
back to periods even before the Bernoulli–Newton episode which
we highlight in this paper.

2. The Challenge Posed by Johann Bernoulli

In essence, the brachistochrone problem posed by Johann Bernoulli

is the following:
Given two points A and B in a vertical plane (Figure 2), what is
the curve traced out by a particle acted on only by gravity, which
starts at A and reaches B in the shortest time?

In Greek, ‘brachistos’ means ‘the shortest’ and ‘chronos’ means Given two points A and
‘time’, hence the name ‘brachistochrone’ of the curve along which B in a vertical plane,
the object traverses in the least time. The history of this problem what is the curve traced
out by a particle acted on
is rather romantic, and is worth visiting. only by gravity, which
starts at A and reaches B
in the shortest time?
Declared Bernoulli:

RESONANCE | September 2017 849

 GENERAL ARTICLE

Figure 2. The solution to

the brachistochrone problem
is not along the shortest dis-
tance, which is of course the
straight line, but along a path
described by a cycloid. This
is now best understood in the
framework of the calculus of
variation.

“I, Johann Bernoulli, address the most brilliant mathematicians in

the world. Nothing is more attractive to intelligent people than an
honest, challenging problem, whose possible solution will bestow
fame and remain as a lasting monument. Following the example
set by Pascal, Fermat, etc., I hope to gain the gratitude of the
whole scientific community by placing before the finest mathe-
maticians of our time a problem which will test their methods
and the strength of their intellect. If someone communicates to
me the solution of the proposed problem, I shall publicly declare
him worthy of praise.”

Bernoulli knew the solution, but he challenged other mathemati-

cians to tackle this problem. Five solutions to this problem are
famous. Those are by (1) Isaac Newton (1642–1727), (2) Jo-
hann’s younger brother Jacob Bernoulli (also known as James or
Jacques Bernoulli) (1655–1705), (3) G W Leibniz (1646−1716),
(4) Guillaume-François Antoine, Marquis de l’Hôpital (1661−1704),
and (5) Johann Bernoulli himself. Newton’s solution was pub-
lished anonymously by the Royal Society (with the help of Charles
Montague), but Johann Bernoulli immediately recognized that it
was Newton’s. He said, “we know the lion by his claw.” New-
ton’s solution (Figure 3a) [7], was essentially the following:

850 RESONANCE | September 2017

 GENERAL ARTICLE

Figure 3. (a) Newton’s re-

sponse to Johann Bernoulli’s
challenge. (b) Marquis de
l’Hôpital’s graphical solu-
tion. (c) Johann Bernoulli’s
solution. The cycloid path
emerges as the limit of the
strips.

From the given point A, let there be drawn an unlimited straight

line APCZ parallel to the horizontal, and on it let there be de-
scribed an arbitrary cycloid AQP meeting the straight line AB
(assumed drawn, and produced if necessary) at the point Q, and
further a second cycloid ADC whose base and height are to the
base and height of the former as AB is to AQ respectively. This
last cycloid will pass through the point B, and it will be that curve
along which a weight, by the force of its gravity, shall descend The principle of
most swiftly from the point A to the point B. extremum action is a
very profound
Jacob Bernoulli then solved it by what we now call as the ‘sepa- mathematical design
ration of variables’ method. which in its various
incarnations explains
Leibniz sent Bernoulli his solution, just a week after the problem why objects move the
was posed. Leibniz provided a trajectory along which the body way they do!
would move, and called it ‘tachystoptota’, which means ‘curve

RESONANCE | September 2017 851

 GENERAL ARTICLE

of quickest descent’. Leibniz provided the correct trajectory, but

did not recognize it as the cycloid. The graphical solution [7] by
Marquis de l’Hôpital (Figure 3b) was published 300 years later
by Jeanne Peiffer in 1988. Johann Bernoulli’s solution (Figure
3c) divided the vertical plane into strips [7]. Bernoulli proposed
that the particle followed a piecewise linear path in each strip.
The problem then reduced to the determination of the angle the
straight line segment in each strip made, and to determine this
Bernoulli invoked Fermat’s principle – that light always follows
the shortest possible time of travel. In the limit, as the strips be-
come infinitely thin, the line segments tend to a curve where at
each point, the angle the line segment made with the vertical, be-
comes the angle the tangent to the curve makes with the vertical.
At the end of the solution, Johann Bernoulli said:
“Before I end, I must voice once more the admiration I feel for
the unexpected identity of Huygens’ tautochrone and my brachis-
tochrone. I consider it especially remarkable that this coincidence
can take place only under the hypothesis of Galileo, so that we
even obtain from this a proof of its correctness. Nature always
tends to act in the simplest way, and so it here lets one curve
serve two different functions, while under any other hypothesis
we should need two curves.”
The principle to which Johann Bernoulli alluded is known after
Pierre Fermat (1601−1665, Figure 4). It is the precursor to the
variational principle. Originally, it explained why a ray of light
takes the path it does (including refraction) when it meets the
boundary of an interface between two media. The principle of
extremum action is a very profound mathematical design which
in its various incarnations explains why objects move the way
they do! For example, it explains why a stream of water running
down a hill would take the path of steepest descent, and also the
trajectories of any mechanical system subject to initial conditions
and governed by an equation of motion that is derivable from the
principle of extremum action.
Rich contributions following Fermat’s work by Maupertuis (1698–
1759) and Euler (1707–1783) culminated in the works of La-

852 RESONANCE | September 2017

 GENERAL ARTICLE

Figure 4. Pierre de Fermat

(1601−1665) was a French
lawyer who pursued math-
ematics as an active hobby.
The principle of extremum
action is named after him.

grange (1736–1813) and Hamilton (1805–1865) which provide a

robust framework for the entire discipline of classical mechanics.

3. The Variational Principle

The variational principle [1–5] rests on the premise similar to

Galileo–Newtonian mechanics that an isolated mechanical state
of the system with one degree of freedom is represented by the
pair (q, q̇), with q representing its position, and q̇ its velocity.
However, the position and velocity in this scheme have a broader
sense which accommodates two distinctive features:
(i) q is the instantaneous coordinate of the object, called its ‘gen-
eralized coordinate’, in the sense that it provides the essential in-
formation about the position of the object under study after ac-
counting for any constraints (if any) which provide partial infor-
mation about the position. The difference between a
generalized coordinate
(ii) q̇ provides the instantaneous velocity, called the ‘generalized and a physical
velocity’. It is the time-derivative of the generalized coordinate. coordinate in
3-dimensional space
q and q̇ need not necessarily have the dimensions L and LT−1 . becomes important when
we consider an
These can be, for example, an angle and angular velocity. Like- N-particle system
wise, the ‘generalized momentum’ is defined as p = ∂L
∂q̇ . It is not subjected to some
necessarily the ‘mass times velocity’ of the Newtonian mechan- constraints.

RESONANCE | September 2017 853

 GENERAL ARTICLE

Nature selects that path ics, although it may well be just that as a possible special case.
for which the action is an The difference between a generalized coordinate and a physical
extremum.
coordinate in 3-dimensional space becomes important when we
consider an N-particle system subjected to some constraints. If
there are m constraints, the number of degrees of freedom re-
duces from 3N to (3N − m), and these are then represented by
(3N − m) generalized coordinates. We refer the readers to some
excellent text books [1, 2], on classical mechanics which detail
the generalized coordinates, velocities, and momenta.
Hamilton’s variational principle states that a mechanical system
evolves in such a way that a mathematical quantity called action
is an extremum; action being defined as the definite integral S
(over time t) of the Lagrangian function L(q(t), q̇(t)) , which itself
will be defined shortly:
 t2
S = L(q(t), q̇(t))dt. (1)
t1

For action to be an extremum, any variation in it when deter-

mined along alternative trajectories, such as those shown in Fig-
ure 1, must be zero; i.e., just as the derivative of a function of
one-variable is zero at an extremum (maximum or minimum). At
these points, the function is ‘stationary’ and any variation in it is
zero. We therefore ask the following question: what is the varia-
tion or change in action, if we imagine the system to evolve along
alternative paths from the initial time t1 to the end time t2 of the
time-interval under consideration? Hamilton’s principle tells us
that the system evolves along a path for which the variation in
action is zero. Nature selects that path for which the action is an
extremum.
The change in action with respect to (a) the path along which the
system evolves as per the laws of nature, and (b) an alternative
imaginable path, is:
 t2  t2
δS = L(q + δq, q̇ + δq̇)dt − L(q, q̇)dt = 0. (2)
t1 t1

854 RESONANCE | September 2017

 GENERAL ARTICLE

From (2), it follows that:

  ∂L
t2
∂L 
0 = δS = δq + δq̇ dt. (3)
t1 ∂q ∂q̇

Now, δq is an arbitrary change in the path; at each instant one

could think of various alternative paths for each of which δq(t)
would be different. The condition under which (3) would hold
for arbitrary δq is, as shown below in Section 4, that:

∂L d  ∂L 
− = 0. (4)
∂q dt ∂q̇

(4) is known as the Euler–Lagrange equation, or in the present

context only as the Lagrange equation. It is an ‘equation of mo-
tion’, just like the one in Newton’s second law, but for the degree
of freedom q rather than for each physical coordinate. If there
are N generalized coordinates (one for each degree of freedom)
{qi , i = 1, 2, 3, ..., N} , then each qi satisfies the Lagrange equation
corresponding to it, and q in (4) must be replaced by qi .

4. The Euler–Lagrange Equation

Leonhard Euler (1707−1783) is one of the most celebrated

mathematician-physicist of all times, respected for his seminal
contributions to a large number of disciplines, including mechan-
ics, optics, fluid dynamics, and musicology apart from fundamen-
tal mathematics. In order to appreciate the solution to the brachis-
tochrone problem, we visit the variational principle [8], which
provides the basis for (4). Consider a function φγ (ξ) of an inde-
pendent variable ξ. Just how the function φγ (ξ) can be written in Lagrange’s equation is
terms of ξ, however depends on a continuous parameter γ, which an ‘equation of motion’,
therefore appears as subscript. For example, we may write: just like the one in
Newton’s second law,
but for the degree of
φγ (ξ) = φγ=0 (ξ) + γχ(ξ). (5) freedom q rather than for
each physical coordinate.

We now consider another function ψγ of ξ which has a rather

RESONANCE | September 2017 855

 GENERAL ARTICLE

complicated dependence on ξ, but it can be expressed in a simple

manner by expressing it as a combination of dependence on three
terms, (i) direct dependence of ψγ on ξ as some simple function
of ξ, (ii) dependence of ψγ on φγ (ξ) , which in turn of course
depends on ξ with details determined by the parameter γ, and
d
(iii) also a dependence of ψγ on dξ φγ (ξ) . We can then write ψγ ξ
in terms of its dependencies on 3 arguments:
 d 
ψγ = ψγ φγ (ξ), φγ (ξ), ξ . (6)
dξ

The Euler equation is a The third argument of ψγ indicates the explicit dependence on
mathematical condition ξ of ψγ (ξ), normally expressible as some simple function of ξ .
that must be satisfied for
In addition to such an explicit dependence of ψγ on ξ , the first
the integral Iγ to be an
extremum with regard to two arguments in (6) stand for an implicit dependence of ψγ on
d
the variations in its value ξ, through its dependence on φγ (ξ) and dξ φγ (ξ). All the three
for different γ. arguments are required to express the complete dependence of
ψγ on ξ. Leaving out one or the other would limit our capacity
to know ψγ in terms of ξ. However, not all of the 3 arguments
may always be required. We now consider the following definite
integral of ψγ (ξ) over the independent variable ξ :
 ξ2  
d
Iγ = ψγ φγ (ξ), φγ (ξ), ξ dξ. (7)
ξ1 dξ

The Euler equation is a mathematical condition that must be sat-

isfied for the integral Iγ to be an extremum with regard to the vari-
ations in its value for different γ. The parameter γ influences how
φγ (ξ) depends on ξ. The extremum can be either a maximum or
a minimum, or even a saddle point. Specifically, it is the station-
ary property of the extremum that is underscored in this analysis.
Suppose the extremum value of Iγ occur at γ = 0. Any change in
γ therefore results in increasing (or decreasing) the value of Iγ if
the extremum is a minimum (or maximum). As we change γ, the

856 RESONANCE | September 2017

 GENERAL ARTICLE

rate at which the integral Iγ changes is given by:

 ξ2  
∂Iγ ∂ d
= ψγ φγ (ξ), φγ (ξ), ξ dξ, (8a)
∂γ ∂γ ξ1 dξ
 ξ2  

∂Iγ ∂ d
i.e., = ψγ φγ (ξ), φγ (ξ), ξ dξ, (8b)
∂γ ξ1 ∂γ dξ
 ξ2
∂Iγ ∂ψγ ∂φγ ∂ψγ ∂φγ

hence, = +  dξ, (8c)

∂γ ξ1 ∂φγ ∂γ ∂φγ ∂γ
∂φγ
where, φγ = . (8d)
∂ξ

From (5) we have:

∂φγ ∂χ
=γ i.e., φγ = γχ . (9a)
∂ξ ∂ξ

∂φγ ∂ 
Hence, = χ(ξ) and φ (ξ) = χ (ξ). (9b)
∂γ ∂γ γ

Accordingly, using (9b) in (8c), we have:

 ξ2

∂Iγ ∂ψγ ∂ψγ

= χ(ξ) +  χ (ξ) dξ. (10a)
∂γ ξ1 ∂φγ ∂φγ
Now, integrating the second of the above terms by parts, we get:
 ξ2
∂ψ
ξ2  ξ2 d  ∂ψ 

∂Iγ ∂ψγ γ γ
= χ(ξ) dξ + χ(ξ) − χ(ξ) dξ.
∂γ ξ1 ∂φγ ∂φγ ξ1 ξ1 dξ ∂φγ
(10b)
Now, χ(ξ2 ) = 0 = χ(ξ1 ), ξ1 and ξ2 being the end points of the
range of the definite integral, and hence the term in the middle of
(10b) is zero.
 ξ2
∂Iγ ∂ψγ d  ∂ψγ 

Hence, = − χ(ξ)dξ. (10c)

∂γ ξ1 ∂φγ dξ ∂φγ

∂I
Now, to get ∂γγ = 0 when Iγ is an extremum for arbitrary χ(ξ),
the necessary condition would be:
∂ψ d  ∂ψγ 

γ
− = 0. (11)
∂φγ dξ ∂φγ

RESONANCE | September 2017 857

 GENERAL ARTICLE

This is a necessary condition for the integral Iγ to have a sta-

tionary value and is known as the Euler equation. Identifying
{I, ψ, φ, ξ} respectively as {S , L, q, t}, we get (4) of Section 3.

5. The Brachistochrone

We now address the brachistochrone problem mentioned in Sec-

tion 1 using the Euler–Lagrange formalism. Let us take a quick
The shape of the curve look at Figure 2. We assume a mass m to go down under grav-
along which the particle ity along frictionless curves such as those sketched in Figure 2.
would traverse under
Bernoulli’s question was to determine the curve that would en-
gravity in least time can
be parametrized in terms able the object to reach the point B, if m is left under gravity
of the height function y from the point A at zero initial speed, in the least time. We use
instead of the time the method of variational calculus described above to determine
parameter t.
this path. We take the zero of the energy scale to be given by
the energy of the particle at rest at point A. As the particle falls
under gravity, it would pick up kinetic energy, while its potential
energy would become negative, the total energy remaining con-
stant. The potential energy at a point where the particle’s coordi-
nate is y is V(y) = −mgy, and the kinetic energy would therefore
be T (y) = 12 mv2 = −V(y) = mgy. Accordingly, the particle’s

speed at any instant is v = 2gy. The time interval δt taken by
the particle to traverse a tiny, (infinitesimal) distance δs is then:
  2
 δx 1 + δy
δs δx2 + δy2 δx
δt = = =  . (12)
v v 2gy

From the above equation, we see that the shape of the curve along
which the particle would traverse under gravity in least time can
be parametrized in terms of the height function y instead of the
time parameter t. The total time taken for the particle to reach

858 RESONANCE | September 2017

 GENERAL ARTICLE

point B, at which the x-coordinate is xB , is therefore:

 xB  1 +  dy 2   xB   
dx 1 + y2
τ=  dx =  dx, (13a)
0 2gy 0 2gy
 
dy
with y = .
dx
√  xB     xB
1 + y2
hence, ( 2)τ = √ dx = Ψ(y, y )dx, (13b)
0 gy 0

 1 + y2
where, Ψ(y, y ) = √ . (13c)
gy

From the Euler equation (11), and equivalently from (4), the con-
dition that the time τ is an extremum is:
∂Ψ d ∂Ψ
 ∂Ψ 

 d ∂Ψ
0= − y = y − y . (14)
∂y dx ∂y ∂y dx ∂y

d ∂Ψ dy ∂Ψ dy ∂Ψ  ∂Ψ 
Now, [Ψ(y, y )] = + = y + y .
dx ∂y dx ∂y dx ∂y ∂y
(15a)
∂Ψ  dΨ ∂Ψ 
i.e., y = − y . (15b)
∂y dx ∂y

dΨ ∂Ψ  d ∂Ψ
Hence, 0= −  y − y . (16a)
dx ∂y dx ∂y
d ∂Ψ

i.e., 0= Ψ − y  . (16b)
dx ∂y

Essentially, this means that:

 ∂Ψ
Ψ − y  = κ, a constant. (17)
∂y

(17) provides a relationship between the coordinate y and its deriva-

tive y . This relation provides the equation to the trajectory for the
fastest path for the particle to get to point B from point A under

RESONANCE | September 2017 859

 GENERAL ARTICLE

gravity, as detailed below.

Now,

∂Ψ ∂ 1 + y 2 y

=  √ =  . (18a)
∂y ∂y gy gy(1 + y 2 )

Hence we get,
y 2
C
Ψ−  = √ , a constant. (18b)
gy(1 + y 2 ) g

√
Separating the factor g and placing it in the denominator of
(18b) enables us to write subsequent relations in a very conve-
nient form.
1 1
i.e.,  =C or squaring y(1 + y 2 ) = = α.
y(1 + y 2 ) C2
(18c)
 dy 2
i.e., y = α − y.
dx
(18d)

We thus get the following relation between x and y which essen-

tially gives us the trajectory, i.e., the shape of the curve, along
which the time taken for the mass m to traverse under gravity
from point A to B would be the least:

y
(δy) = (δx), (19a)
(α − y)
 
y
i.e., x = dy. (19b)
(α − y)

Using the coordinate system shown in Figure 5, a simple substi-

tution:
α θ
y = (1 − cos θ) = α sin2 (20)
2 2
renders the relationship in a familiar form:

θ α
x = α sin2 ( )dθ = (θ − sin θ) + β, (21)
2 2

860 RESONANCE | September 2017

 GENERAL ARTICLE

Figure 5. The solution to

the brachistochrone problem
turns out to be a cycloid,
described in this figure in
terms of angle θ in terms of
which the (x, y) coordinates
are given by (23).

where β is the constant of integration. From the initial condition

θ = 0 when y = 0, we get β = 0. The other constant to be deter-
mined is α, and this is also easily determined since the path must
pass through point B for which the coordinates are given, say, by
(xA , yB ). Hence, xB = α2 (θb − sin θb ) and yB = α2 (1 − cos θb ). From
these two equations, one can determine the constant α. The equa-
tion to the path of the brachistochrone (i.e., the path along which
the time taken is the least) is therefore given by:
α α
x= (θ − sin θ); y = (1 − cos θ), (22)
2 2
which defines a ‘cycloid’ (Figure 5) with a radius of the circle
given by α2 . A cycloid can be described as the locus of a point on
the rim of a circle, such as bicycle wheel, while the center of the
circle itself traverses along a straight line.
From Figure 5, we see that θ = 0, when P is at the origin. The cir-
cle rolls down through a horizontal distance OS along the x-axis.
This distance is exactly equal to the arc-length PS = rθ, since the
radius of the circle is r. The Cartesian coordinates (x, y) of the
center of the circle C, are obtainable in terms of the parameter θ,
since xC = OS = arc-length PS = rθ, and yC = CS = radius of the

RESONANCE | September 2017 861

 GENERAL ARTICLE

circle = r. The Cartesian coordinates of point P are given by:

xP = OS − PQ = rθ − r sin θ = r(θ − sin θ), (23a)
yP = CS − CQ = r − r cos θ = r(1 − cos θ). (23b)

α
The above equations describe a cycloid with r = 2 .
In Figure 6a is shown a small brachistochrone (cycloid) model,
along with three other slopes which we machined from an acrylic
(C5 O2 H8 )n sheet of thickness 15 mm. To machine the brachis-
tochrone, the curvature of an acrylic sheet was contoured to pro-
duce the cycloid from numerical data generated by a small com-
puter program which gave us 400 (x, y) points as per (23), with
π
the angle 0 ≤ θ ≤ 4π in steps of δθ = 100 . The resulting numerical
data generated the cycloid curve, which was plotted using Mas-
tercam X7 CAD/CAM software. We generated the ‘G-Codes’
to machine the acrylic sheet in CNC (Computerized Numerical
Controller) – VMC (Vertical Milling Center) machine. A 6 mm
carbide flat end-mill cutter was used to machine the acrylic sheet.
Care was taken to use sufficient coolant at the machining area to
avoid the melting of the acrylic sheet and get smooth curve edges.
Since the cycloid curve represents the brachistochrone path, this
model is labeled PB . The width of the track was about 1.45 cm,
and brachistochrone was set in a frame of height 22.9 cm and
breadth 25.6 cm.
For comparison, three additional models, with different curva-
tures (including a straight shortest path) were machined from acrylic
sheets. For bookkeeping, these three additional slopes were
labeled as PL , PS , and PD , to respectively represent the linear
(straight) path between the points A and B, a path shallower than
the brachistochrone, and a path deeper than it. The dimensions
of the four models were machined to enable the release from rest
a metallic object held by an electromagnet (Figure 6b), to come
down under gravity alone, from identical start-points (A) of re-
lease, up to identical end-points (B) of their respective trajectories
(Figures 6a and 6b). The start-time at the release of the metallic
object at point A, and the end-time when the object crosses point
B, were recorded using two photo-gates (Figure 6b) and the time

862 RESONANCE | September 2017

 GENERAL ARTICLE

Figure 6a. The experi-

mental setup (left panel) to
perform the brachistochrone
experiment. The time in-
tervals taken by an object
to come down under gravity
alone from identical points
‘A’ to ‘B’ (right panel) were
measured using two photo-
gates, seen in the left panel
(also seen in Figure 7, be-
low). The experiment was
done on four different paths
– PL , PS , PB , and PD as de-
scribed in the text.
Figure 6b. The brachis-
tochrone model placed for
conducting the experiment
to determine the time-
interval δt to measure the
time taken by a mass m
to traverse under gravity
from point ‘A’ to ‘B’ shown
in Figure 6a. A metallic
object is released from rest
(zero initial speed) by the
interval δt the object took to traverse the four paths PL , PS , PB, electromagnet at point A.
and PD from A to B were measured. The photogates A and B
The experiment was done 50 times, and the results for the time respectively record the start
intervals along the four paths are presented in the table below: and end instants of time
from whose difference the
It is clear from Table 1 that the experimental results are com- interval δt is determined.
pletely in accordance with the conclusions arrived from the vari-
ational principle. The cycloid path PB is the brachistochrone.
We note that in our experiment, the object that traversed the four
paths mentioned above was a tiny metallic ball of diameter ∼13
mm. Since the ball would roll down rather than skid, the rota-
tional kinetic energy would need to be added to the translational
kinetic energy in obtaining (12). Hence, the kinetic energy at

RESONANCE | September 2017 863

 GENERAL ARTICLE

Table 1. The time-interval

δt (recorded by the photo-
Path Time Difference δt to Standard Deviation
gates) taken by the metallic
Traverse From ‘A’ to ‘B’
object released at rest from
PL 0.345 0.0033
point ‘A’ to reach point ‘B’
as shown in Figure 6.
PS 0.295 0.002
PB (cycloid) 0.286 0.0025
PD 0.312 0.0023

a point where the particle’s coordinate is y and potential

 energy
1 I 2
is V(y) = −mgy, would be – T (y) = 2 m + r2 v , instead of
T (y) = 12 mv2 = −V(y) = mgy . The total energy would be con-
  
served and hence the particle’s speed would be v = 2 m+m I gy
r2

instead of v = 2gy . Effectively, this would only scale the ac-
celeration due to gravity g by a dimensionless factor m+m I , where
r2
m is the mass of the ball, I its moment of inertia, and r its radius,
without affecting the primary analysis.
There are other brachistochrone paths in nature; two of which
were machined in our workshops which we call as the ‘brachis-
tochrone carom boards’ and are shown in Figure 7. A striker
hit from one of the two foci of the elliptic carom board passes
through the other. On the parabolic carom board, if the striker
strikes the parabolic reflector along the axis of the parabola, then
after reflection, it passes through its focus, as shown in Figure 7.
The parabolic reflector’s focus had a distance of 14.5 cm for its
shortest distance with the parabolic curvature. The major axis of
the elliptic carom board was 55.8 cm while its minor axis was
33.5 cm.
The results of the carom board experiments are akin to the path
taken by a ray of light, as is well known in the laws of optical re-
flection, which are completely determined by Fermat’s principle,
which is the precursor to the variational principle, as discussed
above.

864 RESONANCE | September 2017

 Figure 7. The brachis-
6. Conclusion tochrone carom boards.
Carom board strikers’ inci-
There are other situations in nature where the optimal path would dent along any line parallel
be the one as determined by the variational principle. If a life- to AQ in the parabolic
guard wishes to run and save a person drowning in a river, it is reflector are reflected back
by the parabolic surface
best that he approaches the bank through a path along which he
through the focus F. In
would reach the accident victim fastest, and this must be opti-
the elliptic reflector, a
mized, considering the fact that the speed at which he would run
striker struck from the focus
on land to approach the bank would be different from the speed
F1 always gets reflected
at which he would swim after jumping in the waters. Clearly, through the second focus F2
the solution would be given by the variational principle, and the regardless of which point it
most rapid path would be given as per Fermat’s principle. It is was aimed at.
reported that when a bunch of ants approach a grain of food and
need to crawl over surfaces on which their speeds are different,
they follow Fermat’s principle, and also that when an eagle dives
to catch a prey, it does so along a cycloid. These phenomena in
nature are curious, but we are not aware of any rigorous analy-
sis of these natural phenomena which seem to be consistent with
the variational principle. We trust that the pedagogical analysis
of the variational principle in this work, and the accompanying
experiments shown in Figures 6 and 7 would aid the students in
learning the founding principles of classical mechanics.

Suggested Reading
[1] Herbert Goldstein, Charles Poole and John Safko , Classical Mechanics, 3rd
Edition, Pearson Education, Inc., 2002.

RESONANCE | September 2017 865