You are on page 1of 39

Instruction Manual

and Experiment Guide

OPTICS SYSTEM 2

4864.21

N.B.:

 Pictures, images and descriptions in this manual may not exactly correspond with the
actual items supplied.
 It is also important to note that the experiments in this manual are, only, suggestions.
They are not meant to indicate the limitation of the equipment, which can be used in
wide range of experiments, depending on the educational requirement of the teacher.

1
2
GENERAL DESCRIPTION:

The Optics kit 2 permits the study of light polarization as well as some aspects of light reflection and
refraction. In fact, the kit contains Hartl apparatus, which allows one to perform many experiments in a bi-
dimensional space, related to the reflection of light on mirrors, and to the refraction through transparent
bodies, within the approximation of the geometric optics. The Hartl apparatus can be used with or without the
optical bench, which is included. The basic theory explored by this kit involves the kernel’s laws of physics. In
minutes students or instructors can set up many different experiments.

LIST OF EXPERIMENTS:

 refraction index of a liquid


 refraction index of a glass
 light reflection of a plane mirror
 light reflection on a deformable mirror
 light refraction in a prism
 light refraction through a converging or a diverging lens
 light refraction through different shaped material
 polarization of light
 rotation of the polarization plane of a sugar solution
 total refraction prism
 colours mixing
 focal length
 malus law
 quantitative photometry

ASSEMBLY INSTRUCTIONS:

 Here is the picture sequence to assemble the Hartl disk

 Below illustrates how to use the mirrored side parts for the ray optics and colour mixing box

 Just reverse the ray optics and colour mixing box to produce a parallel or diverging ray beam

3
 Example of Hartl disk usage with the optical bench to study geometrics optics

4
Index of related topics:

B
Biconcave lens.................................................................................................................................22
Biconvex lens...................................................................................................................................22

C
Colours mixing..................................................................................................................................11
concave mirror..................................................................................................................................16

F
Fermat’s principle........................................................................................................................6; 13
Focus of a lens.................................................................................................................................22

H
Hartl apparatus................................................................................................6; 9; 13; 16; 18; 19; 22

I
inverse square law...........................................................................................................................27

L
Light reflection............................................................................................................................13; 16
Light refraction.........................................................................................................................6; 9; 18

M
Malus law..........................................................................................................................................23

O
Optical activity..................................................................................................................................29

P
photometry........................................................................................................................................27
Plane mirror......................................................................................................................................13
Prism..................................................................................................................................................18

R
Refraction index of a glass...............................................................................................................9
Refraction index of a liquid...............................................................................................................6
Rotatory power.................................................................................................................................29

S
Snell’s law...........................................................................................................................................6

T
Total reflection..................................................................................................................................19
Triangular prism...............................................................................................................................19

5
Experiment 1
RELATED TOPICS:

 Hartl apparatus
 Light refraction
 Refraction index of a liquid
 Snell’s law
 Fermat’s principle

The aim of the experiment is to study the refraction index of a liquid

ITEMS NEEDED:

 Hartl disk
 Refraction index vessel
 Ray optics box with single slit

APPARATUS SETTING:

Dispose the apparatus as shown in the preceding picture. Pour enough liquid (e.g. water) into an half of the
vessel. Power on the Ray optics box with the single slit and obtain a beam that crosses diametrically the
Hartl disk.

THEORY:

A light beam that strikes the plane surface of the optical body undergoes a change in direction according to
Snell’s law
n1 sin ( q1 ) = n2 sin(q 2 )
where ni (i=1,2) are the refraction indexes of the two media and θ i (i=1,2) are the angles between the beam
and the perpendicular to the surface.

Since the refraction index of air at STP is nearly 1 we can write also

6
sin ( i ) = n �
sin(r )
where n is the refraction index of the substance we want investigate, i is the angle of incidence and r is the
angle of refraction.
To deduce Snell’s law it’s possible to use Fermat's principle.
This principle states, in it's simplest form, that light waves of a given frequency, choose the path between two
points which takes the least time to traverse. The most obvious example of this is the passage of light
through a homogeneous medium in which the speed of light doesn't change with position. In this case, the
shortest time is equivalent to the shortest distance between the points, which, as we all know, is a straight
line. Thus, Fermat's principle is consistent with light travelling in a straight line in a homogeneous medium.
The speed of light in a medium with refractive index n is c/n where c is its speed in a vacuum. Thus, the time
required for light to go some distance in such a medium is n times the time light takes to go the same
distance in a vacuum.

Referring to the preceding figure, the time required for light to go from A to B this time becomes function of
the variables h1, h2, y, w, c, n
By finding the minimum time, this results in the condition
sin q I = n �
sin q R
We recognize this result as the Snell's law.

PROCEDURE:

By placing the refraction index vessel (with water at 20°C) at different angles of incidence from the light
beam

7
we get the following experimental data where the last column is calculated using equation :

Angle of incidence [degree] Angle of refraction [degree] Refraction index


10° 7° 1.425
20° 15° 1.321
30° 23° 1.280
40° 30° 1.286
50° 36° 1.303
60° 42° 1.294

So, we can conclude that water at 20°C has a refraction index of nearly 1.3 which is fairly close to the
standard value of 1.33.

8
Experiment 2
RELATED TOPICS:

 Hartl apparatus
 Light refraction
 Refraction index of a glass

The aim of the experiment is to study the refraction index of a glass

ITEMS NEEDED:

 Hartl disk
 Trapezoidal prism
 Ray optics box with single slit

APPARATUS SETTING:

Set up the apparatus as shown in the preceding picture.

THEORY:

The refractive index takes a characteristic value for each homogeneous substance in correspondence with
each wavelength.
The following figure suggests a very simple way for determining the approximate value of the refractive index
of a substance that can be used directly when a block of the substance with the shape of a right
parallelepiped is available:

9
Using the thumbtacks, fix a sheet of paper on the wooden plane and lay the glass block, resting on the wider
faces, at the centre.
With a pencil, draw the outline of the glass block on a piece of paper.
Two pins A and B determine the direction of an incident ray on a face of the glass block. The two pins are
fixed on the working plane so that the straight line passing through them forms with the edge of the block a
determined angle.
Looking beyond the glass block, find the position for which the two pins, seen through the glass block, are
aligned, then fix two other pins O and P to determine this new straight line.
Verify that the two pins are aligned on the straight line determined by the other two, viewed from one side
and from the other of the glass.
Take away the glass and draw the lines as shown in the figure and a circumference of any radius centred in
E.
From the Snell’s law we have that
HM
n=
LN

PROCEDURE:

After having drawn the following lines directly on the support table covered by a sheet of paper

we get for the refractive index HM 1.7 cm and LN1.1 cm.
So the index of refraction is nearly 1.5 corresponding to a crown glass.

10
Experiment 3
RELATED TOPICS:

 Colours mixing

The aim of the experiment is to study the additive colours mixing

ITEMS NEEDED:

 Ray optics box with side mirrored parts


 Coloured filters

APPARATUS SETTING:

Set up the apparatus as shown in the preceding picture with the green, blue and red filter.

THEORY:

A Prism can be used to break apart white light into its component colours. This shows that white light is a
mixture of the projected colours
Although the spectrum shows that white light contains a continuous range of colours, colours from just three
areas of the spectrum could be mixed to form any other colour including white. These three colours of light
are from the red, green, and blue parts of the spectrum.
When the three primary colours of light are mixed, the intensities of the colored light is being added. This can
be seen where the primary colour illumination overlaps. The yellow formed when red light is added to green
light is equal to the illumination of the red and green combined.

11
PROCEDURE:

Just watch on a screen what is the additive composition of the furnished colours filters.

12
Experiment 4
RELATED TOPICS:

 Hartl apparatus
 Light reflection
 Plane mirror
 Fermat’s principle

The aim of the experiment is to study the law of reflection for a plane mirror.

ITEMS NEEDED:

 Hartl disk
 Plane mirror
 Ray optics box with single slit

APPARATUS SETTING:

Set up the apparatus as shown in the preceding picture. Power on the Ray optics box with the single slit and
obtain a beam that crosses diametrically the Hartl disk.

THEORY:

To deduce the reflection law it’s possible to use again the Fermat's principle.
As explained in the preceding experiment, this principle states in it's simplest form, that light waves of a
given frequency traverse the path between two points which takes the least time to traverse. The most
obvious example of this is the passage of light through a homogeneous medium in which the speed of light
doesn't change with position. In this case, the shortest time is equivalent to the shortest distance between
the points, which, as we all know, is a straight line. Thus, Fermat's principle is consistent with light travelling
in a straight line in a homogeneous medium.

13
The preceding figure shows a candidate ray for reflection in which the angles of incidence and reflection are
not equal. The time t required for the light to go from point A to point B is easy found as a function of h1, h2,
y, w and c (velocity of light).
If we find the minimum time by differentiating with respect to y and setting the result to zero, we obtain sinqI =
sinqR or
qI = qR

which is the law of reflection.

PROCEDURE:

By placing the plane mirror at various incidence angles we get

and the reflection law is clearly verified.


If the single slit is substituted with the three slits series, the phenomenon can be observed for three parallel
beams.

14
15
Experiment 5
RELATED TOPICS:

 Hartl apparatus
 Light reflection
 concave mirror

The aim of the experiment is to investigate the light reflection on a concave mirror.

ITEMS NEEDED:

 Hartl apparatus
 Concave mirror
 Slide with slits

APPARATUS SETTING:

Power on the projector and place in front of it the screen with three slits, in order to obtain three parallel rays,
the central of which must cross diametrically the goniometric disc.

THEORY:

When a beam of parallel rays strikes a concave or convex mirror, each ray is reflected in different directions
depending on the angle with which it meets the mirrored surface. In particular, if the parallel beam meets a
concave mirror with circular outline and if the opening of the mirror is not too large, all the reflected rays.
pass through the same point (secondary focus). Indeed, if the beam has the same direction as the axis of
the mirror, the point of convergence (main focus) is on that axis at a distance from the mirror equal to half the
curvature radius.
If, instead, the parallel beam meets a convex circular surface, the reflected rays are divergent and their
extensions pass through a single point (virtual focus) behind the mirror.

PROCEDURE:

Place the deformable mirror on the goniometric disc place. By changing its shape you can realize different
shapes as concave and convex mirror and, in particular, at least approximately, as circular mirror. In all cases
the goniometric disc can be rotated to observe the effect of the mirror as the direction in relationship with the
incident beam varies.

16
17
Experiment 6
RELATED TOPICS:

 Hartl apparatus
 Light refraction
 Prism

The aim of the experiment is to study the light refraction through a plane block and through a prism

ITEMS NEEDED:

 Hartl apparatus
 Trapezoidal prism
 Slide with slits

APPARATUS SETTING:

Power on the projector and place the slide with the slit between it and the disc. The slide should be
perpendicular to the beam, in order to obtain that a ray crosses diametrically the disc.

PROCEDURE:

If the trapezoidal prism is correctly placed on the disc along the path of the ray it is possible to observe the
double bending undergone by the ray in crossing the two plane surfaces of the optical body. The two
bendings balance each other and the beam comes out from the optical body with the same direction it had
when it had entered: the only effect is the parallel movement undergone by the beam, the magnitude of
which depends, as it can be easily verified, on the first angle of incidence.

18
Experiment 7
RELATED TOPICS:

 Hartl apparatus
 Total reflection
 Triangular prism

The aim of the experiment is to study the total reflection phenomenon.

ITEMS NEEDED:

 Hartl apparatus
 Trapezoidal prism
 Slide with slits

APPARATUS SETTING:

Power on the projector and place the screen with three slots between it and the disc. The screen should be
perpendicular to the beam, so that the central ray crosses the disc diametrically.

THEORY:

A prism is said to be at total reflection if the light that enters inside it with an appropriate direction undergoes
at least one total reflection before coming out from the prism: that happens when the light, in its path inside
the prism, meets one of the surfaces of the prism with an angle of incidence larger than the critical angle. .

PROCEDURE:

Put the triangular prism on the disc, along the path of the parallel rays beam, as shown in the following
figures:

19
20
 In the first case (Fig. 1), thanks to the special symmetry of the system, the refracted rays come out
from the prism with an angle (with respect to the face perpendicular) equal to the first incidence
angle. In particular, if the incident rays are parallel to the face on which the total reflection happens,
the rays that come out maintain the original direction.
 In the second case (Amici’s prism) no deviation appears on the two equal faces of the prism and the
rays are bent only on the third face, by which they are totally reflected.
The beam comes out bent 90° from the original direction (Fig. 2)
 In the third case (Porro’s prism) the rays enter and exit perpendicularly to the hypotenuse of the
prism, while they are bent twice, for total reflection, on the other faces. The beam that comes out
maintains the original direction, but its direction of propagation results will be reversed (Fig. 3)

21
Experiment 8
RELATED TOPICS:

 Hartl apparatus
 Biconcave lens
 Biconvex lens
 Focus of a lens

The aim of the experiment is to study the refraction of a parallel rays beam through a converging or diverging
lens

ITEMS NEEDED:

 Hartl apparatus
 Biconvex lens
 Biconcave lens
 Slide with slits

APPARATUS SETTING:

Power on the projector and place the screen with three slots along the beam and perpendicularly to it, as
shown in the figures.

PROCEDURE:

Fix on the disc, along the path of the beam and perpendicularly to it, one converging lens (biconvex) in order
to observe the different bending undergone by the three rays and the existence of a focal point. By taking a
graph of the converging rays, it is possible to find the focal length both for converging and diverging lenses.
Substitute the converging lens with the diverging one (biconcave) to observe the different behavior of this
last one and the existence of a virtual focus.

22
Experiment 9
RELATED TOPICS:

 Malus law

Aim of the experiment is to explain and verify the Malus law.

ITEMS NEEDED:

 Ray optics box


 Pair of polarizing filters
 Photoresistance
 Digital multimeter (not included in the kit)

THEORY:

Polarized filters are most commonly made of a chemical film applied to a transparent plastic or glass surface.
The chemical compound used will typically be composed of molecules that naturally align in parallel relation
to one another. When applied uniformly to the lens, the molecules create a microscopic filter that absorbs
any light matching their alignment.
Let A be the amplitude of plane polarized light incident on a polarizer. Vibration direction of polarizer is
inclined at angle θ to direction of polarization.

The component of A, A·cosθ is transmitted and A·sinθ is blocked out.


Recall that the intensity is proportional to the square of amplitude, so that the intensity transmitted is:

cos 2 q
IT = I �
where I is the intensity of light incident on polarizer. This is called Malus law.

23
2p
�I u cos 2 q dq I u ) is
If unpolarized light Iu falls on the polarizer, ideally only half of the light (i.e. < It >=0
=
2p 2
transmitted.
If a second polarizer is now placed in the beam, with vibration direction θ relative to the first polarizer, the
cos 2 q
Iu �
intensity transmitted, according to Malus law, is IT = and varies from 0 to Iu/2.
2
PROCEDURE:

Prepare the apparatus as shown in the preceding picture. The converging lens is not strictly necessary. By
means of the digital multimeter take a reading of the resistance in kOhm of the photoresistance every 10°
between the first polaroid and the second. Then convert this resistance value by taking it’s inverse since the
conductivity of the photoresistance is inversely proportional to its resistance and the conductivity is directly
proportional to the light intensity.

PHOTORESISTANCE:

The photoresistance is called also a L(ight) D(ependent) R(esistor). LDR are made from cadmium sulphide,
a material which, when prepared properly, contains no or very few free electrons when kept in complete
darkness. Its resistance is therefore quite high. When it absorbs light, electrons are liberated and thus the
material becomes more conducting. Cadmium sulphide is therefore called a photoconductor. The electrons
are free only for a limited time and when the light is switched off, they are captured again by those places
where they originally came from and thus the conductor turns again into an insulator.

Let us consider a disk of cadmium sulphide provided with two electrodes

The distance between the electrodes is d and the length is l. When the disk is exposed to an illumination L a
number of electrons N are liberated per second in the disk between the two electrodes

N =h L l d

where h is a constant depending on the wave length of the light. When a voltage V is applied to the
electrodes the electrons move with a velocity proportional to the field strength E
V
v = mE = m
d
The proportionality constant m is called the mobility. Not all electrons may reach the positive electrode but
only those which are liberated within a distance v· where  is the average life time of a free electron.
The fraction of the electrons that contribute to the current is therefore
v
d
and the measured photocurrent i is from , and given by
v  h L l e mV 
i = Ne =
d d
where e is the electric charge of the electron.
The resistance R, caused by the illumination, is then

24
V d
R= = L-1
i h e m l
The life time  is usually not constant but depends on the wave length  of the light and on the illumination L
 =  0 (  ) L- b
The relation between the resistance and the illumination can therefore be expressed in good approximation
by
A
R=
La
where R is expressed in Ohm, L in lux and a is a constant. By taking the logarithm of both sides equation
we get
log10 R = log10 A - a log10 L
and this relationship is often given by the manufacturer in a logarithmic scale:

The value of a depends e.g. on the cadmium sulphide used and the manufacturing process. Values around
0.7-0.9 are quite normal

EXPERIMENTAL DATA:

Here are some experimental data showed with the best fit interpolating curve. It’s evident how the
experimental points are well explained with a cos 2 dependence. Moreover the slight difference in phase
means that the start angle must be corrected in order to have the maximum intensity at 0°.

25
Experiment 10
RELATED TOPICS:

 photometry
 inverse square law

Aim of the experiment is to compare the power of illumination of two different kind of luminous sources in a
quantitative way.

ITEMS NEEDED:

 Ray optics box


 Photoresistance
 Digital multimeter (not included in the kit)

A punctual light source (as the filament of the ray optics box) of luminous intensity I (Candela/cd) emits a
light flux  (Lumen/lm) throughout a solid angle . Luminous intensity in a solid angle element d amounts
to:
d
I=
d
If an area dA is illuminated by a luminous flux d, the luminance E (Lux/lx) is:
d d d  1
E= = � = I �2
dA d dA r
This equation describes the photometric law of distance. According to this, the luminance E of a surface
decreases proportionally to the square of distance r for constant luminous intensity I.

PROCEDURE:

Prepare the apparatus as shown in the preceding picture. By means of the digital multimeter take a reading
of the resistance in Ohm of the photoresistance every 5 cm. Then convert this resistance value by taking it’s
inverse since the conductivity of the photoresistance is inversely proportional to its resistance and the
conductivity is directly proportional to the luminance of the source.

26
EXPERIMENTAL DATA:

Here are some experimental data showed with the best fit interpolating curve. It’s evident how the
experimental points are well explained with a 1/x2 dependence.

27
Experiment 11
RELATED TOPICS:

 Optical activity
 Rotatory power

The aim of the experiment is to show how some substances such as sugar react when they are crossed by a
beam of polarized light: they rotate the polarization plane of the light around its direction of propagation
(optical activity).

ITEMS NEEDED:

 6 sliders for holders and 1 slider for projector


 Ray optics box
 Power supply
 Pair of polarizing filter
 Cylindrical Plexiglas tanks
 Biconvex lens +50 with lens-holder (optional but suggested)
 Biconvex lens + 100 with lens-holder (optional but suggested)
 Coloured filters
 White screen
 Sugar and/or dextrose and/or fructose

The optical activity is a phenomenon connected with the "asymmetry by reflection" of the molecules of many
substances. Let us consider an asymmetric molecule by reflection (the kind of asymmetry between the two
hands). To be more precise, let that molecule have a helical shape. The polarized light that arrives with the
electric vector oscillating along the helix axis (z) sets in motion the electrons.

28
The electron motion along z generates a second wave (coming out) that is linearly polarized according to z.
However, since the electrons are compelled to move along the helix, some components of the motion will be
perpendicular to the planes passing through z (see figure).
A pair of such components would send (for instance in P) waves in phase opposition (that, therefore, would
annul each other) if there would not be the time lag due to the fact that only one wave must cover the
diameter d of the helix. It is therefore necessary to correct the wave that comes out and that was initially
considered polarized along z with a small field component polarized along the x axis (see figure).
The wave that will actually come out has the polarization plane rotated of an angle q with regards to the
plane of the incident wave, even when the linearly polarized light strikes a set of asymmetric molecules
arranged at random (for instance in a water solution). In fact each molecule appears identical if seen from
one side and from the opposite.
The amount of rotation is determined by the rotatory power of the optically active substance present in the
solution and by the amount of molecules of the substance that interact with the beam of light. The amount of
molecules is determined by the degree of concentration of the solution and from the path length of the light in
the solution.
There are some variables that can affect the rotation: temperature and wavelength of the light. Since the
rotation changes according to the colour of the light, we will first use a white light and then we will continue
the experiment with coloured filters. All the measurements obtained will be recorded in the tables you find
later.
In polarimetric measurements the specific rotatory power of a substance is defined as follows:

q
PT =
L� c

where PT stands as the specific rotatory power at a given temperature T and wavelength ; (we will not
consider  when using white light). q stands as the angle of rotation of the polarization plane of the light. L
stands as the path length of the light in the substance (in our case it will coincide with the length of the
container filled with the solution). c stands as the concentration of the solution expressed in grams of solute
(grams of solvent; in our experiments the only solvent we will use is water).

If you wish to use the equipment to determine the concentration of a given solution (but in order to do so you
will have to calibrate it) all you have to do is to invert the above mentioned formula
It is rather clear that the formula allows one to determine the concentration by determining the width of the
rotation angle of the light.

COMMENTS:

A few reminders about some sugars: Glucose (C 6H12O6) is also known as dextrose because solutions of
glucose rotate the plane of polarized light to the right. Glucose occurs in ripe fruits, the nectar of flowers,
leaves, saps, and blood and is variously called starch sugar, blood sugar, grape sugar, and corn sugar.
Fructose (also C6H12O6, differs from glucose in the location of the doubly bonded oxygen atom) is also known
as levulose because it behaves oppositely to glucose in that it rotates polarized light to the left. Fructose is
also known as fruit sugar since it is in many fruits and vegetables. It is the major sugar in most types of

29
honey because nectar contains a high proportion of fructose). Sucrose (C 12H22O11) is our common table
sugar. Sucrose is abundant in sugar cane and sugar beet, which are its principal sources. It is a molecule
consisting of two glucose-like units; one unit is a six-member glucose ring and the other is a five-member
fructose ring.
When the enzyme invertase is added to a sucrose solution, the two parts of the sugar molecule are
separated, giving a solution of equal parts of glucose and fructose known as invert sugar, the name
stemming from the fact that it rotates polarized light in the opposite direction of sucrose.

HOW TO MAKE THE EXPERIMENT

First assess the environmental temperature; then assemble the equipment according to the instructions
given in the following picture

Fill the tank with 100 g of water; switch on the projector and place the polaroid filters according to the
instructions given in Table 1; make sure that the axis of the filters are vertically aligned at 90°; then position
the tank at such a height to allow the image of the surface of the water to be clearly seen; the image should
be visualized on the lower half of the luminous circle.

In order to make sure the image projected on the screen is luminous and clear, follow the assembly
instructions given in the table below

30
Now add 10g of sugar in the tank and stir the solution until the sugar is completely dissolved in the solution.
Let the solution stand still for a couple of seconds, then rotate the axis of the Polaroid analyser until the lower
half of the image on the screen is totally obscured. Then record the width on the angle indicated on the
optical disk of the Polaroid analyser filter. This angle is the angle of rotation of the polarization plane of the
light.

Fig. 5

Repeat the experiment: first place the coloured filters in a row; then record the width of the angles. (see
Figures 6 and 7)

Fig. 6

31
Fig.7

Now, increase the quantity of sugar in the tank and repeat the measurements; a 40% concentration of sugar
is the maximum concentration allowed for these experiments. Record the values you obtained in the table
below:

q° q° q° q° q°
L (cm) c % white P yellow P red P green P blue P
10 10
15
20
25
30
35
40
(P: rotatory power)

Now use a longer tank (15 cm) repeat the test and record the results in the
table below.
q° q° q° q° q°
L (cm) c % white P yellow P red P green P blue P
15 10
15
20
25
30
35
40

Then replace the 15 cm long tank with the 20 cm long tank and do the test again:

q° q° q° q° q°
L (cm) c % white P yellow P red P green P blue P
20 10
15
20
25

32
30
35
40
The processing of the data obtained (provided you consider experimental errors and the possibility that the
substance you used, sugar, is not 100% pure) allows the operator to verify the Kernel’s law which we
illustrated at the beginning of this instructions manual. The law establishes a proportional relationship
between the specific rotatory power of a substance mixed in a solution at a given temperature, its
concentration, and the path length of light in the solution. The extent of the rotatory power is determined by
the wavelength of light.

33
APPENDIX
Basics of experimental error theory

We can say that everything we known about the physical world has an inherent uncertainty. In
particular, when we experimentally investigate something there is always an “experimental error”
and an “experimental precision”. Since one of the main features of experiments is their
reproducibility, it is very important to deal with this subject in order to be able to explain how good
our results are. This is possible with “experimental error theory”, a scientific approach to this
problem.

Let us consider the following example: find the density of a solid rubber cube.

o First trial (with very raw instrument). We can estimate that the mass of the cube is
nearly 50 g and the length of a side is nearly 6 cm. So the density would be:
M M
r= = 3 = 0,23148... . There are many questions: “Where can I stop with
V L
decimal digits to communicate my result?” “Is it better to have precision on the
mass measurement or on length measurement?” “How do we combine our
experimental error on the mass measurement with the experimental error on length
measurement?”
o Second trial (with more accurate instruments). By using an electronic balance and
a meter stick I find a mass of 60g and a side length of 5,4 cm. So the density would
be:
M M
r= = 3 = 0,381039475... . We still need to answer the questions posed above
V L
but we also have to answer a new question: “What makes this trial better than the
first one?”
o Third trial (with much more accurate instruments to measure the side length). If
we improve the accuracy of the length measurement, for example by using a vernier
caliper, the problem becomes more involved. This is due to the fact that we do not
get the same result if we make more than one measurement. Instead we have a set of
different measurements like (5,455 cm; 5,425 cm; 5,465 cm; …). So we are again
faced with the question: “Which one of the measurements (5,455 cm; 5,425 cm;
5,465 cm; …) do should I take?”

Therefore, the more we analyse the problem the more it gets involved. To search for a possible
solution we can start from the third trial and observe that, generally speaking, when we improve the
accuracy of an instrument we reach a point at which the experimental results are not unique but are
scattered around some value as illustrated in this graph:

34
If the number of measurements N is greater then about 30, the distribution of the experimental data
is bell-shaped and has a value X for which there is a maximum and around which the data are
scattered in a nearly symmetrical way. It is also possible to distinguish a value that determines an
interval around X into which a significant percentage of the measurements are placed. We need to
answer the questions: “Is X the best estimate of our measurement?”, “How much can we rely on
this value?” and “What percentage of the measurements are in the interval X-s and X+s ?”

To express these questions mathematically, we could try a prototype function that fits our data and
that expresses the probability to get a particular measurement value:
1

0.8

0.6

0.4

0.2

-3 -2 -1 1 2 3
this is the graphical representation of the function
2
f ( x) = e - x
If we want to centre the function around the value X we use the expression x-X in place of x, and if
we want to control the scattering of the measurements around X it’s possible to divide (x-X)2 by
2s 2.
The following figure shows f(x) with X=2 and s=1;1.5;2
1 1 1

0.8 0.8 0.8

0.6 0.6 0.6

0.4 0.4 0.4

0.2 0.2 0.2

-2 2 4 6 8 10 -2 2 4 6 8 10 -2 2 4 6 8 10

35
Finally, if we want to control the area under the curve we have to multiply it by a normalization
factor As that would depend on s .
Therefore our prototype function is:
( x- X ) 2
-
f ( x) = As �
e s2
2�

where X is the value for which we have the maximum and s determines how the measurements are
scattered around X. This is called a “Gaussian function” or a “Normal function”, but the underlying
data represent a distribution (still called Gaussian) and not a function. It can be proved that the
Gaussian distribution is derivable from the binomial distribution assuming that the number of
measurements N à ¥ and s remains constant.

The physical meaning of all this is that we do not describe a measurement with a single number but
rather with a set of values each one with its own probability to appear as an experimental datum.
This probability is governed by the “Gaussian distribution”. There is an analogy with quantum
mechanics (for example with the wave packet of a particle) where the interpretation is that if we
make a measure of the position of the particle then the probability to obtain a particular value is
governed by the Gaussian function and is never a well defined fixed value.

Let us determine the value of As in . We must have a probability of 1 to get a measurement in the
range from -¥ to +¥ (that is, if we perform a measurement we are certain to get some kind of result
no matter how large or how small that result is):
+� 1

-�
f ( x )dx = 1 � As =
2p �s

To give an interpretation of s we can ask what happens if we are only interested in the probability
of finding measurements in the range from X-s to X+s instead of the range from -¥ to +¥:
2
t
X +s 1 +1 -
�X -s
f ( x) dx =
2p �
� e 2 dt : 0.68
-1

so s , also called then “standard deviation” (s 2 is called “variance”), is the amount of uncertainty we
have to allow for, in the most probable value X, if we want to claim a roughly 68% chance of
correctly predicting the result of any single measurement.

To determine X, also called the “mean value”, we consider a set of N measurements x1, x2, …, xN.
The probability to get a single result between xi and xi+dx is:

( x - xi ) 2
1 -
Pi = �
e s2
2�

s 2p
so the probability to get all the results (viewed as independent events) is:
N

�( x - xi )
2

1 - i=1 2�s 2
P = P1 �P2 �... �PN a N�e
s
Since we are speaking about the probability P to get all the results and we can suppose to have
already done our experiment with a set of real results what should be the value of P?

36
If we accept the maximum likelihood principle we can make an analogy with entropy and say that P
is proportional to the entropy obtained from our experiment. The value X must be a point of
maximum entropy. By the second principle of thermodynamics we have to maximize P, otherwise
d �N 2�
said X is the value of x that minimize the exponent: �� ( x - xi ) �= 0 from which it results:
dx �i =1 �
1 N
�xi
X =
N i =1
that is, the mean value X is the arithmetic mean and describes all the collected data since it is the
value for which the maximum entropy is obtained for our set of data.

� �( x - xi ) �
N
2

d �1 - 2�s 2 �
i =1

To determine s we can proceed in the same way �N �


e �= 0 from which:
ds �
s �

� �

1 N
� ( x - xi )
2
s=
N i =1
But what should be use instead of x in equation ? If we use then equation is slightly self-referential
2
1 N �1 �
because s = � � ( x1 + ... + xi + ... + xN ) - xi � and the i-esim term appears two times. It is
N i =1 �N �
possible to show that the correct value of the standard deviation is:

1 N
� ( X - xi )
2
s=
N - 1 i =1
Clearly s is not defined for N=1 (we are assuming N greater of nearly 30, otherwise there are better
distributions to consider).

Suppose now we have a function Q of several variables Q = f (a, b, c,...) and we want to know how
the experimental error on each variable contributes to Q.
We can say that by varying the variables, the quantity Q varies of:
�Q � Q �
Q
DQ = Da + Db + Dc + ...
�a �b �c
and if we identify our uncertainty Dx with the standard deviation s x we can say that:
� Q �Q � Q
sQ = sa + sb + s c + ...
�a �b � c
the modulus is due to the fact that errors could cancel each other and we want to consider the
maximum error.
We could do better, obtaining a smaller value, if the variables are normal and independent, by
1 N
� ( �- Qi ) where Qi = f (ai , bi , ci ,...) is the i-esim value of Q by taking
2
starting from s Q =
N - 1 i =1
the i-esim value of each variable of our set of data, �= f ( A, B, C ,...) is the mean value of Q by
taking the mean value of each variable of our set of data.
2 2 2
��Q �Q � �� Q� �� Q�
( �- Qi ) = ( DQi ) = � Dai + Dbi + ... �; � �( Dai ) + � �( Dbi ) + ...
2 2 2 2
Since (by
��ai �bi � �� ai � �� bi �
neglecting terms of higher order) we have:
37
2 2
��
Q� 1 N ��
Q� 1 N
� ( ) � ( Dbi ) + ...
2 2
sQ = � � D ai + � �
��a �1
N4- 14 i2=1 4 43 �� b �1
N4- 14 i2=1 4 43
s a2 s b2

or
2 2
�� Q� ��
Q�
s Q = � �s a2 + � �s b2 + ...
�� a� ��b�
which is better of since it’s always lower.
N
1
Suppose now that the function Q is just the arithmetic mean X =
N
�x
i =1
i . By applying equation
we get
2 2
��
X� ��
X�
s X = � �s x21 + � �s x22 + ...
��
x1 � ��
x2 �

X ��1 N
� 1
but

=
xi �xi �
�N
�x �= and s
� N
i =1
i x1 = s x2 = ... = s and so

s
sX =
N
which is called “standard deviation of the mean”. Analogously to the “standard deviation”, it tells us
how good is the mean value X and we can assume it as the amount of uncertainty we have to allow
for if we want to claim a roughly 68% chance of correctly predicting the result of any other mean
value it is possible to obtain.
sQ
It is also useful to speak about relative error instead of absolute error s Q . The relative error can
Q
be expressed in percentage.

For example let us return to the problem of determine the density of a cube.
M
Now, the function Q is the density which is function of the mass M and the side length L: r = .
L3
60 g g
If M = 60 g and L = 54mm it’s easy to find that the mean value is r = 3 3
10-4 3
= 3,81 �
54 cm cm
By applying equation we have that the relative error is:
sr 1 � r 1 � r L3 1 L3 -3M 1 3
= sM + sL = s M + sL = sM + sL .
r r � M r � L M L 3
M L 4
M L
If we can suppose the precision of the mass measurement is s M = 2 g and the precision of the
length measurement is s L = 1mm we have:
s r 2 3� 1
= + ; 3,3% + 5, 6% = 8,9%
r 60 54
(this says it is more important to make a careful length measurement than a careful mass
measurement).
By applying equation we get a better (lower) estimate of the density error:
sr sM 2 9� s L2 22 9 �12
= + = + = 6, 5% .
r M2 L2 602 542

38
This means that if we take another measurement of density there’s a probability of nearly 68% that
g
the new value will lie between (3,8 �0, 2) � 10 -4 3 .
cm
It is important to note that since the standard deviation on density is 0,2 x 10-4 g/cm3 we can stop at
the first decimal digit 3,8 x 10-4.

39

You might also like