PHYSICS LAB

EXPERIMENTS
LICENSE, DISCLAIMER OF LIABILITY, AND LIMITED WARRANTY

By purchasing or using this book (the “Work”), you agree that this license
grants permission to use the contents contained herein, but does not give you
the right of ownership to any of the textual content in the book or ownership
to any of the information or products contained in it. This license does not
permit uploading of the Work onto the Internet or on a network (of any kind)
without the written consent of the Publisher. Duplication or dissemination
of any text, code, simulations, images, etc. contained herein is limited to and
subject to licensing terms for the respective products, and permission must
be obtained from the Publisher or the owner of the content, etc., in order to
reproduce or network any portion of the textual material (in any media) that
is contained in the Work.

MERCURY LEARNING AND INFORMATION (“MLI” or “the Publisher”) and anyone

involved in the creation, writing, or production of the companion disc,
accompanying algorithms, code, or computer programs (“the software”),
and any accompanying Web site or software of the Work, cannot and do
not warrant the performance or results that might be obtained by using
the contents of the Work. The author, developers, and the Publisher have
used their best efforts to insure the accuracy and functionality of the textual
material and/or programs contained in this package; we, however, make
no warranty of any kind, express or implied, regarding the performance
of these contents or programs. The Work is sold “as is” without warranty
(except for defective materials used in manufacturing the book or due to
faulty workmanship).

The author, developers, and the publisher of any accompanying content,

and anyone involved in the composition, production, and manufacturing of
this work will not be liable for damages of any kind arising out of the use of
(or the inability to use) the algorithms, source code, computer programs,
or textual material contained in this publication. This includes, but is
not limited to, loss of revenue or profit, or other incidental, physical, or
consequential damages arising out of the use of this Work.

The sole remedy in the event of a claim of any kind is expressly limited to
replacement of the book, and only at the discretion of the Publisher. The
use of “implied warranty” and certain “exclusions” vary from state to state,
and might not apply to the purchaser of this product.
PHYSICS LAB
EXPERIMENTS

Matthew M. J. French

MERCURY LEARNING AND INFORMATION

Dulles, Virginia
Boston, Massachusetts
New Delhi
Reprint and Revision Copyright ©2017 by MERCURY LEARNING AND INFORMATION
LLC. All rights reserved.

Original title and copyright: Finding Your Way in the Undergraduate Physics Labo-
ratory. Copyright ©2013 by M.M.J. French. ISBN 978-0-9549780-8-2. All rights
reserved. Published by The Pantaneto Press.

This publication, portions of it, or any accompanying software may not be repro-
duced in any way, stored in a retrieval system of any type, or transmitted by any
means, media, electronic display or mechanical display, including, but not limited
to, photocopy, recording, Internet postings, or scanning, without prior permission
in writing from the publisher.

Publisher: David Pallai

MERCURY LEARNING AND INFORMATION
22841 Quicksilver Drive
Dulles, VA 20166
info@merclearning.com
www.merclearning.com
1-800-232-0223

M.M.J. French. Physics Lab Experiments.

ISBN: 978-1-942270-80-5

The publisher recognizes and respects all marks used by companies, manufactur-
ers, and developers as a means to distinguish their products. All brand names and
product names mentioned in this book are trademarks or service marks of their
respective companies. Any omission or misuse (of any kind) of service marks or
trademarks, etc. is not an attempt to infringe on the property of others.

Library of Congress Control Number: 2016935948

161718321 Printed in the United States of America

This book is printed on acid-free paper.

Our titles are available for adoption, license, or bulk purchase by institutions, corpo-
rations, etc. For additional information, please contact the Customer Service Dept.
at 800-232-0223(toll free).

All of our titles are available in digital format at authorcloudware.com and other
digital vendors. The sole obligation of MERCURY LEARNING AND INFORMATION to the
purchaser is to replace the book, based on defective materials or faulty workman-
ship, but not based on the operation or functionality of the product.
 CONTENTS

About the Author ix

List of Figures xi
List of Tables xv

Chapter 1 Introduction 1
1.1 Characteristics of the Laboratory 2
1.2 Demonstrating Undergraduate Physics Laboratory 4

Chapter 2 Safety 9
2.1 Hazard Symbols 10
2.2 Electrical 12
2.3 High Voltages 13
2.4 Gas Cylinders 15
2.5 Vacuum Pumps 19
2.5.1 Pressure Units 20
2.5.2 Hand Pumps and Vacu-vins 22
2.5.3 Diaphragm Pump 22
2.5.4 Rotary Pump 22
2.5.5 Diffusion Pump 23
2.5.6 Turbomolecular Pump 23
2.6 Cryogenics 25
2.6.1 Liquid Nitrogen 25
2.6.2 Dry Ice 30
2.6.3 Liquid Helium 32
2.6.4 Helium-3 33
2.6.5 Liquid Oxygen 35
2.6.6 Argon 36
2.6.7 Cryogen Free Cooling 37
2.7 Lasers 38
2.8 Ionizing Radiation 40

Chapter 3 Making Measurements 45

3.1 SI Units 46
3.2 SI Prefixes 47
3.3 Vernier Calipers 47
3.4 Digital Calipers 48
3.5 Micrometer 48
3.6 Balances 49
vi • Physics Lab Experiments

3.7 Ammeters 50
3.8 Voltmeters 51
3.9 Resistor Color Codes 52
3.10 Multimeters 52
3.11 Further Ideas on Measurement of Resistance 53
3.12 Constant Current Sources 55
3.13 Lock-in Amplifers 58
3.14 Further Ideas on Reducing Noise in Electrical Measurements 61
3.15 Signal Generator 64
3.15.1 TTL 64
3.16 Oscilloscopes 65
3.17 Light Gates 70
3.17.1 Velocity with One Light Gate 70
3.17.2 Velocity with Two Light Gates 71
3.17.3 Acceleration with Two Light Gates 71
3.17.4 Acceleration with One Light Gate 71
3.18 PASCO Science Workshop Interface 72
3.19 Data Harvest 74
3.20 Pressure Gauges 77
3.21 Temperature Measurement 78
3.21.1 Near Room Temperature 79
3.21.2 Low Temperatures 82
3.21.3 High Temperatures 84
3.22 PID and Temperature Control 84
3.23 Kaye and Laby Tables 86
3.24 Microscopes 87
3.24.1 Optical Microscopes 87
3.24.2 Transmission Electron Microscope 89
3.24.3 Other Microscopes 90
3.25 Spectroscopes 92
3.26 Computers 92
3.26.1 Plugs, Sockets, and Connectors 92
3.26.2 Computer Data Logging 97
3.26.3 Still Cameras and Scanners 105
3.27 Video Cameras 106
3.28 Other Researcher's Data 107
3.29 Wii Controllers 108
3.30 Traditional Photographs 109

Chapter 4 Data Analysis and Errors 113

4.1 Accuracy and Precision 115
4.2 Measurement Errors 117
 Contents • vii

4.2.1 Random Error 117

4.2.2 Systematic Error 117
4.2.3 Zero Error 117
4.3 Uncertainty 118
4.4 Resolution 118
4.5 Tolerance 118
4.6 Sensitivity 119
4.7 Response Time 119
4.8 The Mean of the Sample 120
4.9 Variance and Standard Deviation of the Sample 121
4.10 Measurement Distributions 123
4.11 Standard Error on a Single Measurement 126
4.12 Standard Error on the Mean 126
4.13 Estimation of Errors 127
4.14 Gaussian Error Distribution 128
4.15 Combination or Propagation of Errors 129
4.15.1 Linear 130
4.15.2 Products and Quotients 131
4.15.3 General Functions 133
4.15.4 Appropriate Significant Figures in Final
Answers and Errors 134
4.16 Plotting Appropriate Graphs 134
4.16.1 Drawing Lines of Best Fit and Finding Gradients 135
4.16.2 Error Bars on Graphs 135
4.16.3 Straight Line Graphs 136
4.16.4 Log Graphs 137
4.17 Graph Drawing Software 137
4.17.1 Excel 2003 and Prior Versions 138
4.17.2 Excel 2010 140
4.17.3 Kaleidagraph 144
4.17.4 MATLAB 147
4.18 Least Squares Fitting 151
4.18.1 Regression line of y on x 151
4.18.2 Steepest Descent 154
4.18.3 Gauss-Newton 155
4.18.4 Levenberg-Marquardt Method 158
4.18.5 Nelder-Mead Method 160
4.18.6 Distribution Testing 164
4.19 Uncertainty in Gradients and Intercepts of Graphs 165
4.20 Fourier Transform 168
4.20.1 Window Functions 171
viii • Physics Lab Experiments

Chapter 5 Presenting Results 173

5.1 Lab Book 174
5.2 Reports 175
5.2.1 Software 175
5.2.2 Structure 177
5.2.3 Graphs, Tables, and Diagrams 179
5.2.4 Referencing 183
5.2.5 Equations 185
5.2.6 Special Characters 189
5.3 Posters 192
5.3.1 Software 192
5.3.2 Paper Sizes 193
5.4 Presentations 194
5.4.1 Software 200

Chapter 6 Common Experiments 201

6.1 Example Experiments 201
6.1.1 Measuring Acceleration due to Gravity 202
6.1.2 Measuring the Speed of Light 208
6.2 Example Reports 215
6.2.1 Example Report: Charge Mass Ratio of an Electron 216
6.2.2 Example Report: Investigating Radioactivity 225
6.2.3 Example Report: Latent Heat of Vaporization
of Liquid Nitrogen 251

Index 275
 ABOUT THE AUTHOR

Dr. Matthew French studied for an undergraduate physics degree

at the University of Bristol. While working there towards a PhD
in experimental condensed matter physics, he spent four years
demonstrating experiments in the first year undergraduate physics
laboratory. Now an experienced school teacher and Head of Physics
at Prior Park College, Bath (UK), he is well positioned to bridge the
gap between students' knowledge at the end of secondary school and
that expected of them at the beginning of their university courses.
 LIST OF FIGURES

1. The undergraduate laboratories at the University of Oxford.

Left: condensed matter laboratory. Right: optics laboratory. 3
2. The undergraduate laboratories at the University of Bristol 3
3. Examples of simple low EMF power supplies.
Usually the yellow outputs give AC.
The black (negative) and red (positive) outputs give DC 13
4. Left: Low (electric) potential difference connectors
with exposed metal contacts. Right: High (electric)
potential difference connectors with shielded metal contacts.
The shielding automatically moves back as the plug is connected
into a socket. Both leads can be piggy-backed to connect
multiple leads to the same place. 14
5. Examples of high EMF (EHT) power supplies 14
6. A bracket and belt anchoring a cylinder to the desk 15
7. A cart used to move gas cylinder. This cart
has an extra set of wheels so the user doesn't
have to support part of the weight of the cylinder
while moving the cylinder 16
8. A gas regulator. This one is for oxygen only: the knob is
labeled and the crossed out symbol of an oil can is
present in the pressure gauges 17
9. Left: A gas cylinder key. Right: The spindle on a gas bottle 18
10. A vacu-vin hand pump 21
11. A rotary pump 21
12. A turbomolecular pump backed with a rotary pump. 24
13. Phase diagram for Nitrogen. Nitrogen is a gas at STP
(standard temperature and pressure) 26
14. A neoprene plastic bucket for storing liquid nitrogen. 27
15. Liquid Nitrogen Dewars. Left: a 25 liter “onion” Dewar
with a cart Right: a 10 liter Dewar 28
16. Phase diagram for Carbon Dioxide. Carbon Dioxide is
a gas at STP (standard temperature and pressure). 31
17. Phase diagram for Helium-4. Helium-4 is a gas at STP
(standard temperature and pressure) 32
18. Phase diagram for Oxygen. Oxygen is a gas at STP
(standard temperature and pressure) 36
xii • Physics Lab Experiments

19. Phase diagram for Argon. Argon is a gas at STP

(standard temperature and pressure) 37
20. Left panel: a lead lined wooden storage box and
forceps. Right panel: inside the box and the source inside
a metal cup holder 42
21. A Vernier Scale. The top is the main scale and the
bottom in the traveling scale. The left panel shows a
reading of 0.0 mm; The middle panel shows a
reading of 0.1mm; The right panel shows a reading of 0.6 mm. 47
22. A digital calipers measuring; Left: the diameter of a wire and
Right: the inside diameter of a pipe 48
23. A micrometer measuring a: 5.21 mm and b: 3.76 mm. 49
24. Two different multimeters 53
25. Simple diagram of a four-wire method of measuring the
sample resistance. R1, R2, R3, and R4 are resistances
which represent the contact and lead resistances 54
26. Simple four-wire mounting configuration of a sample to
measure resistance along in the direction of the current flow 55
27. Left panel shows a circuit layout of a resistor (typically
Rseries = 1k to 10 k) used as a constant current source.
It is connected in series with the sample and the oscillator
output on a lock-in amplifier. Right panel shows a circuit
diagram of a (electric) potential difference controlled constant
current source. The opamp used is an Linear Technology
LTC1150CN8#PBF (RS order number 5455629) 57
28. The front panel of a Stanford Research Systems (SRS)
830 lock-in amplifier 59
29. Left panel shows the circuit setup for a normal 4-wire
measurement configuration. Right panel shows the circuit
setup when setting the phase. Current is passed though the
sample and a series resistor, R. The (electric) potential
difference drop is measured across only the resistor 61
30. Inductive cross-talk between wires in two closed
circuits carrying AC signals 62
31. A signal generator 65
32. An CRT oscilloscope 66
33. A digital oscilloscope. 69
34. A PASCO Science WorkShop 500 Interface box
connected to a photogate 72
35. Choosing a sensor in PASCO DataStudio 73
36. Using PASCO DataStudio to measure the velocity in a photogate 74
37. Using PASCO DataStudio to produce a distance time graph 75
 List of Figures • xiii

38. A Data Harvest QAdvanced Datalogger with a

Light Level probe connected to input 75
39. A screen shot of the Data Harvest EasySense software
showing a graph of Light Level against time 76
40. Illustration of chromatic aberation of a lens. Red light
(the dashed line) is refracted less than blue light (the dotted line) 88
41. Illustration of spherical aberation of a lens 88
42. Left: Electron diffraction pattern from Gold. Right:
A dark field image of islands of Gold 90
43. Creating a new VI in LabVIEW 102
44. The LabVIEW program. On the left is the “block
diagram” window and on the right is the “front panel” window 103
45. Inserting a “GPIB Read” component in LabVIEW. 103
46. A LabVIEW program which reads data and produces a graph 104
47. A LabVIEW program within a “while” loop 105
48. Photograph of a graticule using a microscope 106
49. Using ScanIt to set the axis 108
50. Targets illustrating the differences between the terms
accurate and precise. The center of the target corresponds
to the true value. The stars are the individual measurements 115
51. Graphs illustrating the differences between the terms
accurate and precise. The vertical line corresponds to the
true value. The vertical axis is a probability distribution
(or the number of measurements) and horizontal axis is
the magnitude of the quantity being measured 116
52. Number lines comparing the spread of data 123
53. Histogram of the measurements of the current flowing
through a bulb given in the text 124
54. Histogram of the measurements of a quantity when the
number of measurements tends to infinity 124
55. Plotting an “XY Scatter” graph with Excel 2003 139
56. Selecting the data series with Excel 2003 139
57. Adding a polynomial trend line of order 2 to the data
and ensuring the equation is displayed on the chart. 140
58. Adding X axis error bars. They have a fixed size of ± 0.25.139 141
59. Plotting a graph with Excel 2010 141
60. Selecting the data series with Excel 2010 142
61. Adding axis titles with Excel 2010 142
62. Adding a linear trend line to the data and ensuring the
equation is displayed on the chart 143
63. Adding X axis error bars. They have a fixed size of ± 1. 143
64. Switching between editing X and Y error bars 144
65. Plotting a graph with Kaleidagraph 145
66. Selecting data for a graph with Kaleidagraph 145
xiv • Physics Lab Experiments

67. Adding a Trendline with Kaleidagraph 146

68. The fit parameters box in Kaleidagraph 146
69. Adding error bars in Kaleidagraph 147
70. A simple MATLAB graph 148
71. A MATLAB graph showing two lines and axes labels. 149
72. A MATLAB graph with vertical error bars 150
73. An Excel trendline, showing the equation of the straightline
and the R2 value are given by Equations 84, 85, and 86 154
74. The different triangles involved in a two parameter version of the
Nelder-Mead function minimization algorithm. 161
75. A graph showing the maximum and minimum gradients
consistent with the error bars 166
76. The Microsoft Excel 2003 Regression Window 167
77. The Microsoft Excel 2003 Regression package output. 168
78. Entering the data ranges into Fourier Analysis window. 169
79. Filling in the FFT Frequency column. 170
80. Graph of FFT Frequency against FFT Magnitude. 171
81. Fast Fourier transform of a sine wave with frequency
10 Hz. Left panel shows leakage with no windowing function
present. Right panel shows leakage reduced
with a windowing function 172
82. Good and bad scaling of axes 179
83. Good and bad lines on graphs 180
84. Always add x and y error bars where possible 180
85. The Equation Tool Bar in Microsoft Word 2003 185
86. Inserting an Equation in Microsoft Word 2010 186
87. The Equation Tool Bar in Microsoft Word 2010 187
88. Inserting Greek characters in Microsoft Word 2010. 187
89. Inserting an Equation in Open Office Writer 189
90. Example poster 1. Based on a final year project 195
91. Example poster 2. Based on early work towards a PhD. 196
92. Example presentation slides 199
93. Moment of inertia of a pendulum 206
94. Faucault method diagram 210
95. Reflection from the fixed mirror as the rotating
mirror changes angle 211
96. Considering the virtual image of the fixed mirror 212
 LIST OF TABLES

1. Chemical Hazard Symbols. Usually the border is red

and the diagram is black 11
2. Non-Chemical Hazard Symbols. Usually the text, border
and diagram is black and the background is yellow. 12
3. Compressed gas cylinder colors 19
4. Conversion factors for common pressure units 20
5. Table of temperature and vapor pressures for He-4 34
6. Table of temperature and vapor pressures for He-3 35
7. The new laser classifications. 40
8. The old laser classifications 42
9. List of common radioactive sources. Protactinium
and Radon are listed with radiation given out by the
decay chain which generates them from Uranium-238
and Thorium-232 respectively 43
10. The SI units 46
11. The SI prefixes 46
12. Resistor color codes 51
13. Thermocouples, Compositions, Temperature
Ranges and Sensitivities 81
14. Effect of changing PID values 86
15. Relationships for setting PID values 86
16. Common plugs and sockets 93
17. Example data for calculation of the mean and variance. 121
18. Probability that a given measurement is within “z”
standard deviations of the mean. 129
19. Example data for a regression line 152
20. Example calculations for a regression line 153
21. Example data for Nelder-Mead method 163
22. Commonly used LaTeX symbols 191
23. Papers Sizes 193
 CHAPTER

1
INTRODUCTION

For some students, “the experimentalists”, their laboratory

sessions are the highlight of their undergraduate degrees. It is in
the laboratory that the physics they have learnt in lectures begins to
make sense to them and the connections between the equations and
reality are illuminated. Some of these students will continue work-
ing in laboratories to complete doctorates in experimental physics
or turn to experimental work in industry. For other students, “the
theoreticians”, the laboratory sessions are something that must be
endured, often on a weekly basis. However, separating students into
the discrete boxes of experimentalists and theoreticians does not
really convey the holistic experience today’s undergraduates receive:
they must be talented at experiments as well as theory in order to
excel at university.
This new book aims to guide both the “experimentalist” and “the-
oretician” through their compulsory laboratory course forming part
of an undergraduate physics degree. Some students will have left
school believing that physics experiments “never work”. Designing
and performing good physics experiments is a craft, it is a skill devel-
oped gradually over time. I hope this book might go some way to
persuading all students of the value and beauty within a carefully
planned and executed experiment and help them develop the skills
to carry out experiments themselves.
Those students who are completing undergraduate research
projects and beginning a doctorate will also find helpful and relevant
sections. Graduate students supervising laboratory sessions will find
2 • Physics Lab Experiments

this a useful resource: especially the dedicated section on “Demon-

strating Undergraduate Physics Laboratory”.
Chapter 2 includes, the characteristics and safety consider-
ations of a range of equipment and materials. Chapter 3 moves on to
detail how many common measurement instruments are operated.
In Chapter 4, the focus turns to how to analyze measurements and
Chapter 5 deals with the presentation of results. Finally, Chapter 6
details a number of common experiments and example laboratory
reports.
The idea of learning about experimental physics through read-
ing a book, might at first seem to be tinged with a hint of irony.
However, there are many concepts and techniques which a phys-
ics student needs to learn and understand. Often these are not
specifically written down or taught in lectures: it is up to the inter-
ested student to investigate and learn for themselves. While the
goal of this book is not to detract from this inquisitive and curiosity
driven learning, it aims to provide an overview of the many topics
and skills required in the first year undergraduate physics labora-
tory and beyond. Inevitably, some readers will find detail lacking
in one area or another or pages which are currently superfluous to
their needs. This book should not be read cover to cover, rather
it should be dipped into to gain a basic understanding of a certain
topic when necessary. Where needed the reader, then armed with
a good grounding in the topic, can use other books or the Internet
for further research.

1.1 CHARACTERISTICS OF THE LABORATORY

In the first year, students will usually work in pairs (or very
small groups) undertaking a different experiment each week. It
is important students are able to work with other scientists: for
some experiments it is essential to have two people to perform the
experiment. By discussing experimental methods and the prob-
lems experienced with their partner students can challenge their
understanding and learn from each other. However, it is essential
to work together as a team and that both are engaged in each part
 Introduction • 3

of the experiment. Each student should record their own results

and perform their own analysis. Of course, the calculations and
results after each stage in the analysis should be cross checked.
Identifying each other’s mistakes soon after they are made will
work to the advantage of both students.

Figure 1.1: The undergraduate laboratories at the University of Oxford. Left: condensed
matter laboratory. Right: optics laboratory.

Figure 1.2: The undergraduate laboratories at the University of Bristol.

Laboratory sessions will probably consist of a day, or most of a day,

to perform the experiment and collect results. Students may have to
hand in their analysis and write up of the experiment the same day,
4 • Physics Lab Experiments

or be given a number of days to complete this. Usually a number of

the experiments are written up as formal laboratory reports, this is
discussed in more detail in Section 5.2.
When entering the laboratory, the experiments will usually be
partially set up on tables in a large room: see for example Figure
1.1 and Figure 1.2. There will probably be other students nearby
performing the same experiment using an identical set of apparatus.
This provides more opportunities to discuss results and troubleshoot
the equipment and the method when results don’t seem to make
sense. Each group of 8 or 10 students (4 or 5 pairs) will usually be
supervised by a post graduate laboratory demonstrator. When advice
or help is needed, having carefully thought about the question and
tried to find the solution the student should consult the demonstra-
tor for help. Sometimes, the demonstrator will remain with the same
group of students for the whole year working with them each week
on different experiments. In other cases a demonstrator will remain
with the experiment and work with a different group of students
each week.
Never touch or remove equipment from someone else’s experi-
ment. Experiments can be working and recording data even if the
user is not present. It is very annoying and frustrating if an experi-
ment is carefully setup, recording data and someone disturbs or
removes some of the equipment.
In the second year, and perhaps third year laboratory courses
experiments become more complex and longer: perhaps 4, 6, or 10
days (usually at 1 day per week). Students still work in pairs, but the
scripts become less detailed and it is expected that students use the
knowledge and skills acquired in the first year laboratory to develop
a method and derive much of the analysis themselves.

1.2 DEMONSTRATING UNDERGRADUATE

PHYSICS LABORATORY

This section is written as a brief introduction for anyone dem-

onstrating undergraduate physics laboratory: usually this would be
concurrent with studying for a PhD.
 Introduction • 5

Demonstrating undergraduate physics laboratory can be an

enjoyable experience which also provides a little extra money. It is
also good experience for the future when applying for post-doc jobs
or academic posts. It is usually expected that all academics working
at a university contribute to teaching in some way. When spending
much of the time working on the computer reading papers or ana-
lyzing data, a day in the undergraduate laboratory can give a very
welcome break. At other times, when experiments are going well
and significant progress is being made the interruption caused by
the laboratory day can be a little frustrating.
In some universities or year groups the demonstrator is attached
to a group of around 8–10 students who work in pairs. Each week
a different experiment is performed and it is the demonstrators’
responsibility to ensure they know how to carry out the experiment
according the the script given to the students, use all the necessary
equipment and complete the analysis. Demonstrators should be on
hand with their groups of students for the full duration of the labora-
tory session. At least initially, this can seem like a great deal of work
and a little daunting for anyone not confident with the experiments.
However, there is usually the opportunity to practice the experi-
ments when the laboratory is closed to undergraduate students and
more experienced laboratory demonstrators who have worked for a
number of years are usually happy to help.
Alternatively, in some universities or year groups the demon-
strator is attached to a particular experiment (or a small number of
experiments). This allows the demonstrator to become an expert in
certain experiments, although this means they meet different stu-
dents each week.
The laboratory books of the students are usually handed to the
demonstrator for checking every week. The demonstrator will be
responsible for marking the work and submitting the marks for colla-
tion. It is good practice for each demonstrator to keep a record of all
marks they award. It is wise to always keep on top of marking. It need
not take significant amounts of time, especially if it is done as soon as
possible after the books are handed when it will be more fresh in the
mind. Calculations and error analysis need not be worked through for
each student individually and checked in detail, although sometimes
it is easy to spot and highlight/correct an error the student has made.
6 • Physics Lab Experiments

Usually some guidance is provided to help with the assignment

of marks, but the following could be used as a guide if nothing is
provided. Bear in mind that a 2:2 would correspond to 5/10, a 2:1 to
6/10 and a 1st to 7/10.
A mark of 4/10 or less would be awarded for a write up which
did not meet the criteria for a mark of 5/10. This will generally be
awarded rarely.
A mark of 5/10 would correspond to a reasonable attempt to
collect data in the laboratory, perhaps incomplete due to a lack of
effort, skill/patience or hindered by temperamental equipment.
There should be an attempt to describe the method in some detail
and to complete the analysis as far as a final answer. The final answer
may be significantly incorrect due to errors in the recoding of data or
the analysis. There may be no consideration of uncertainty.
A mark of 6/10 would be awarded where there was a complete or
almost complete set of data, perhaps only limited by temperamental
equipment. The majority of the steps in the method are described
in the correct order. The analysis has been completed and a final
answer has been calculated. The final answer should be order of
magnitude correct, or there are one or two simple errors which can
be easily identified such as a missing constant, an incorrect conver-
sion or a power of ten error. There may some consideration of uncer-
tainty in the final result but a reasonable, quantitative estimate is not
given.
A mark of 7/10 would correspond to a collection of a complete
set of data of an appropriate accuracy and precision. All the key
points in the method would be described in the correct order and
in significant detail. The analysis should be complete through to the
final answer. The final answer should be order of magnitude correct
or any larger discrepancy identified and an explanation attempted.
There is some consideration of uncertainty leading to a quantitative
estimate of the uncertainty in the final answer, although its origin
(both in the experiment and in numerical terms) may be somewhat
unclear.
A mark of 8/10 or higher will probably only be awarded rarely.
This would correspond to meeting and exceeding all the criteria
for a mark of 7/10. There should be clearly recorded data which
has been repeated a number of times (as time allowed) along with
 Introduction • 7

a detailed consideration of the uncertainty in each measurement.

The method should provide significant detail, which would allow a
similar student to perform the same experiment without reference
to other texts. The final answer should be quoted with an uncer-
tainty which is clearly calculated and well founded in the actual mea-
surements taken.
Alternative guidance is sometimes provided for marking formal
reports. Marks might be awarded in categories such as “Clarity of
arguments”, “Scientific Content”, “Presentation”, “Scientific Liter-
acy and Style”.
The “Clarity of arguments” section might score 7/10 if there is a
clear introduction, explanation of the method, the analysis and the
results. This should be at a level understandable by a student in the
same academic year as the author. If some of these areas are lacking
a clear explanation then 6/10 may be more appropriate. If most of
these areas are not clearly explained then 5/10 or less may be appro-
priate.
The “Scientific Content” section might score 7/10 if there is suf-
ficient detail on all aspects of any derivations needed, the method,
analysis and error analysis which would allow a similar student to
repeat the work without reference to other texts. Extremely thor-
ough and competent work would score  8/10, and work where some
sections are not fully addressed would score 6/10. If many sections
are not fully addressed or if any are missing a mark of  5/10 would
be appropriate.
In the “Presentation” section, marks are awarded for the layout
and organization of the work. Tables and diagrams should be neat,
appropriate and fully labeled with column headings, units and figure
captions (e.g., Figure 1.2: A diagram of ...). Graphs should be of an
appropriate type (usually scatter graphs, without the points joined
up) have axis labels, units, a line of best fit (in most cases), a figure
caption and be large enough to be seen clearly. The written work
should be logically ordered and there should be a good use of sec-
tion numbers and sub-headings. Where there are only a few minor
mistakes a mark of 7/10 should be awarded. A mark of 6/10 would
be appropriate if there are a number of mistakes or one point is con-
sistently absent. If there are many of the above points absent then a
mark of  5/10 could be awarded.
8 • Physics Lab Experiments

Finally marks in the “Scientific Literacy and Style” are awarded

for the style in which the report is written. Usually reports should be
in a formal style with no use of the words “we”, “I” or “you”. Descrip-
tion of the physics involved and the analysis should make correct use
of appropriate technical terms. Reference should be made to other
work which is cited in an appropriate, formal style. If these criteria
are mostly met then a mark of 7/10 should be awarded. If there are
occasional mistakes, then 6/10 could be given. With many mistakes
or a whole report written in the first person (using “we”, “I” or “you”)
then a mark of 5/10 could be awarded.
 CHAPTER

2
SAFETY

While performing experiments at school students probably took

it for granted that the equipment they used was safe and was very
unlikely to do them harm unless used in an obviously unsafe manner.
The class teacher would have been following carefully thought out
advice from a national organization called CLEAPSS [1], in addi-
tion to performing their own risk assessment. In most cases com-
mon sense and following the teachers’ instructions was all that was
required on the student’s part.
At university, students are likely to meet liquid gases, danger-
ous chemicals, high temperatures, high (electric) potential differ-
ences and lasers. Risk assessments should have been carried out for
all instructed experiments met in the first two or three years of an
undergraduate degree. However, some of the equipment or chem-
icals used will have the potential to cause serious harm. Further-
more, other equipment or chemicals which may be present in the
same area could also be very dangerous.
Students should make sure risk assessments are read and
understood and follow any safety rules or advice given by the
university: students have legal obligation to do this in the United
States. Students should begin to see why certain precautions are
needed and develop responsibility for themselves. Equipment
or chemicals which students do not understand or are not con-
fident with should never be touched or used. Where one person
may be confident, another may not be: students should always
ask for advice if they are unsure or not confident with certain
equipment.
10 • Physics Lab Experiments

There is general safety advice which is applicable to all laborato-

ries at all time. Later, safety advice for specific pieces of equipment
will be covered.
Shoes should always be worn when working in the lab because
they provide some protection from spillages and falling objects.
Sandals, high heels, and open shoes are completely inappropriate
as they allow chemical or cryogenic liquids (used by the students
themselves or others) to pose a danger to the feet.
Loose or dangling clothing, jewelery, hair, and beards may get
caught in moving equipment: especially motors. Sensible clothing
should be worn, jewelery removed and hair tied back using a hair
band or elastic band.
If working with chemicals, clothing should ideally cover legs and
arms and a lab coat should be worm to help protect against spills or
splashes.
Food and drinks should never be present or consumed in the
lab. There is the risk of contamination of the food from chemicals,
including traces of unknown chemicals from previous work which
took place in the same area. Drinks can easily be spilled causing
damage to expensive equipment.

2.1 HAZARD SYMBOLS

The Globally Harmonized System of Classification and Labeling

of Chemicals (GHS) uses the pictograms shown in Table 2.1. There
is not officially a single word meaning for any of the symbols: there
should be adjacent written hazard information. However, there is a
general guide given here based on the old European descriptions of
similar hazard symbols.
 Safety • 11

Table 2.1: Chemical Hazard Symbols. Usually the border is red and the diagram is black.

Further non-chemical hazard symbols which might be present

in a laboratory are shown in Table 2.2. The symbols may be shown
with or without the accompanying text.
12 • Physics Lab Experiments

Table 2.2: Non-Chemical Hazard Symbols. Usually the text, border, and diagram is black
and the background is yellow.

2.2 ELECTRICAL

All devices provided in the lab powered by mains electricity

should have been tested prior to use by the university to ensure
they are safe. This is called a portable appliance test (PAT) and
is carried out at set time intervals depending on the equipment.
It involves a visual examination of the wire and plug and a resis-
tance measurement to ensure that there is a good earth connec-
tion from any exposed metal on the case. Because it was deemed
safe at the time of testing, does not guarantee it is safe at the time
of use.
Before plugging in any device to the mains it is prudent to visu-
ally check the cable and plug for any cracks, cuts, breaks, or exposed
wires. Any equipment which is found damaged or not working
should not be used. It should be reported to a lab demonstrator or
lab technician who will remove the equipment or label it as unser-
viceable. Never use equipment marked as unserviceable or U/S or
which has had the mains plug cut off.
 Safety • 13

Many devices operate from mains power or have a built in trans-

former and/or AC-DC converter. Additionally mains powered, power
supplies may be used. There are two major categories of power
supplies which adapt the standard mains supply (UK and Europe:
230 V at 50 Hz and USA: 115 V 60 Hz) to power other devices: low
EMF power supplies or extra high tension (EHT) supplies.
Devices which require a particularly large amount of power may
be connected to a special “three-phase” socket.
Low EMF power supplies are classed as those supplying up to
28 V AC or 40 V DC. At this (electric) potential difference, exposed
connectors can be used as this (electric) potential difference is gen-
erally classed as safe to touch. Examples of low EMF power sup-
plies are shown in Figure 2.1 and suitable exposed connectors are
shown in Figure 2.2. Hand built digital electronic circuits and most
apparatus will comfortably fall within the limit of a low EMF power
supply. Working with EMFs above these limits should be avoided if
at all possible.

2.3 HIGH VOLTAGES

Where there is a scientific need to use (electric) potential dif-

ferences higher than 28 V AC or 40 V DC, great care must be taken
to avoid anyone touching exposed contacts. If the device is mains
powered, this should have been built into the design of the device.
Sometimes an EHT power supply will be needed which may deliver
in the region of 5000 V (5 kV) or more. Special connection cables

Figure 2.1: Examples of simple low EMF power supplies. Usually the yellow outputs give AC.
The black (negative) and red (positive) outputs give DC.
14 • Physics Lab Experiments

Figure 2.2: Left: Low (electric) potential difference connectors with exposed metal contacts.
Right: High (electric) potential difference connectors with shielded metal contacts. The
shielding automatically moves back as the plug is connected into a socket. Both leads can
be piggy-backed to connect multiple leads to the same place.

Figure 2.3: Examples of high EMF (EHT) power supplies.

should be used which have shielded connectors to reduce the risk of

electrocution: see Figure 2.2.
No matter what form of power supply is connected to a device
it is always safest to turn off the supply before touching any metal
or “live” component. Special care should be taken with EHT power
supplies: they must not simply be switched off using a power switch
(on the power supply or plug socket). The (electric) potential dif-
ference should be slowly decreased to zero and a voltmeter (or volt-
age display built in to the power supply) should be used to ensure
the potential difference has reached zero before the power supply is
switched off. The risk of switching off an EHT incorrectly is the pos-
sibility that circuit elements or components (most likely a capacitor)
 Safety • 15

may still retain significant separated charge. This charge creates a

high potential difference which can cause a dangerously large cur-
rent to flow through you when any exposed wiring is touched (even
if the power supply is switched off and unplugged from the mains).

2.4 GAS CYLINDERS

Cylinders of gas provide a reliable and consistent supply of pure,

dry gas to an experiment. Common gases used in university experi-
ments include oxygen, carbon dioxide, nitrogen, helium, argon, and
hydrogen.
The use and movement of gas cylinders can be potentially dan-
gerous, although it can be done safely if the risks and dangers are
understood.
Gas cylinders should always be stored and used in an upright
position. They must always be secured so as they can’t fall over:

Figure 2.4: A bracket and belt anchoring a cylinder to the desk.

16 • Physics Lab Experiments

usually in a transport cart or using a chain or belt which securely

attaches the cylinder to a wall or table bracket: see Figures 2.4 and
2.5. Cylinders which are not secured may fall over and cause a crush-
ing injury to someone nearby. A further severe danger is that the
regulator may be sheared off, allowing gas to escape rapidly turning
the cylinder into a torpedo capable of major injury and penetrating
concrete walls [2].
Generally, cylinders not in use will be stored outside in a cage to
keep them secure. Great care should be taken if a gas cylinder must

Figure 2.5: A cart used to move gas cylinder. This cart has an extra set of wheels so the user
doesn’t have to support part of the weight of the cylinder while moving the cylinder.
 Safety • 17

be moved. Cylinders are heavy and have no “carrying handles”. They

should be strapped securely into a transport cart: see Figure 2.5.
Care should be taken when moving carts as they usually need to
tipped and part of the weight of the cylinder supported by the user’s
arms. Carts should be taken slowly and carefully over uneven floors
and doors should be propped open in advance.
Gas cylinders store gas under high pressure, commonly this is
around 230 bar. To reduce this pressure for use in an experiment a
regulator must be used. The regulator is attached to the outlet at the
top of the cylinder. Regulators for hydrogen and other flammable
fuel gases have a left handed thread on the nut, whereas regulators
for oxygen and inert gases have a right handed thread. Special care
should be taken to ensure that regulators for oxygen cylinders are
labeled as oil free: usually with a crossed out symbol of an oil can.
This reduces the risk of fire.
To fit a regulator to a cylinder begin by ensuring the regulator
is turned off by turning the regulator knob counterclockwise. Next

Figure 2.6: A gas regulator. This one is for oxygen only: the knob is labeled and the crossed
out symbol of an oil can is present in the pressure gauges.
18 • Physics Lab Experiments

check the screw threads on the cylinder and regulator are in good
condition. With closed eyes, blow into the cylinder head to remove
any dust.
Then place the regulator above the thread and using a spanner,
turn the nut (clockwise from above for right handed threads) until
the nut will not turn any further. Do not use excessive force.
To obtain gas, ensure the knob is turned fully counterclockwise (the
off position). Use the cylinder key to turn the spindle (see Figure 2.7)
fully counterclockwise (the on position). The pressure gauge now gives
a reading, indicating the contents of the cylinder. As gas is used the pres-
sure reading will decrease: an empty cylinder will read approximately
zero when the spindle is open. A quick check for leaks in the regulator
can be made. Gas leaking rapidly will make a hissing noise. Slowly turn
the regulator knob clockwise (to the on position) to obtain the gas. The
second gauge indicates the delivery pressure of the gas. The further the
knob is turned clockwise the higher the delivery pressure.
To close down a gas cylinder when the experiment is finished
turn the regulator knob fully counterclockwise to the off position.
Use the cylinder key to move the spindle fully clockwise to the
off position.
Next turn the knob clockwise to remove any gas in the regulator.
This prevents damage to the diaphragm in the regulator. The pres-
sure gauge will now read zero. Finally turn the knob fully counter-
clockwise again.
To remove the regulator when the cylinder is empty and needs
changing follow the instructions in the paragraph above to close
down a gas cylinder. Then unscrew (counterclockwise from above

Figure 2.7: Left: A gas cylinder key. Right: The spindle on a gas bottle.
 Safety • 19

for a right handed thread) the nut using a spanner. The regulator
should now be free. It is vital that the spindle is fully turned off
(turned fully clockwise) before starting to unscrew the nut. This pre-
vents gas escaping rapidly from the cylinder potentially causing the
torpedo effect described earlier.
Note carefully that the spindle is off when turned fully clockwise
where as the regulator is off when turned fully counterclockwise.
Compressed gases are graded by their purity which is usually
expressed as a percentage, for example: 99 % or 99.999 %. How-
ever this is often re-expressed in terms of the number of nines:
99 % becomes 2N (corresponding to 2 nines) and 99.999 % becomes
5N (corresponding to 5 nines). A further variation is 99.95 % which
becomes 3N5 (corresponding to 3 nines and then a five).
In the EU, the colors of cylinders of some gases are becoming
standardized following the new standard EN 1089-3. Colors for cyl-
inders of the most common gases are given in Table 2.3.

2.5 VACUUM PUMPS

Vacuum pumps are used to remove air or other gases from a

sealed container. Depending on the pressure required different
types of pump are used. Commonly used pumps include a simple
wine-saver hand pump, a rotary pump, a diaphragm pump, a diffu-
sion pump, or a turbo pump.

Table 2.3: Compressed gas cylinder colors.

20 • Physics Lab Experiments

The primary danger when using a vacuum pump to reduce the

pressure inside a container is that the container walls fail and the
container implodes. Wall pieces may then rebound off each other
and fly outwards causing injury. It is usual for glass or plastic pres-
sure vessels (such as bell jars) to be surrounded by a mesh cage or
a safety screen to provide some protection in the event of an implo-
sion.
The magnitude of the force acting on a container under vacuum
can be demonstrated with a used soft drinks can. A small amount of
water is placed in the bottom of the empty can which is then heated.
Once the water is boiling and the can contains steam, the can is then
rapidly turned upside down with the opening submerged in water
to seal the can. The water rapidly cools and the steam condenses
reducing the pressure in the can causing it to fold inwards. A reduc-
tion of pressure of around 30 mbar (to around 970 mbar) in the can
is sufficient to cause it to implode.

2.5.1 PRESSURE UNITS

Most pumps or vacuum systems will have a pressure gauge to

measure the pressure achieved. The gauge may read in a variety of
units so it is important to be aware of how to inter-convert units. The
SI unit of pressure is the N/m2 derived from the equation P=F/A.
1 N/m2 is identical to 1 Pascal, abbreviated to 1 Pa. Other common
units for pressure include the atmosphere (atm), the Torr (which is
equivalent to mm of Mercury) and the pound per square inch (psi).
Conversion factors between these units are shown in Table 2.4.

Table 2.4: Conversion factors for common pressure units.

 Safety • 21

Figure 2.8: A vacu-vin hand pump.

Figure 2.9: A rotary pump.

22 • Physics Lab Experiments

2.5.2 HAND PUMPS AND VACU-VINS

A hand pump or vacu-vin is primarily designed for removing air

from bottles of wine, thus preserving an opened bottle of wine for
longer (see Figure 2.8). When the appropriate pressure is reached
this is indicated by the device with a “click”. Only rarely do these
have a built in pressure gauge, but typically they will achieve a pres-
sure of around 400 mbar before the “click”. Inside the vacu-vin a
piston compresses air from the vessel being evacuated and forces it
out through a valve.

2.5.3 DIAPHRAGM PUMP

Diaphragm pumps consist of a vibrating diaphragm which com-

presses air from the experiment and forces it out through a valve.
Typically they achieve a pressure of around 20 mbar.

2.5.4 ROTARY PUMP

Rotary pumps are the work horse pump of the physics lab. They
provide a cheap, reliable way to achieve a vacuum down to a pres-
sure of around 0.1 mbar. A two stage rotary pump consists of two
rotary pumps in series and will produce a vacuum of around 0.001
mbar = 1×10−3 mbar. Figure 2.9 shows a rotary pump.
The pump consists of a vane projecting from a cylinder rotating
eccentrically which compresses air and forces it out through a valve.
Rotary pumps have oil inside. This is a special type of oil which has a
very low vapor pressure i.e., it produces very little vapor. If it did not
have a low vapor pressure it could not be used as the oil vapor would
pollute the vacuum the pump was trying to produce. Despite the
low vapor pressure tiny droplets (i.e., a mist) of oil from the pump
can be carried away by the air it exhausts. This is more of a concern
when the pumping flow rate is high or the pump is run for pro-
longed periods of time or in a confined space/small room. There are
 Safety • 23

statutory limits on the amount of oil mist permissable in air and so

rotary pumps are usually fitted with an oil mist filter.
Some pumps have an anti-suck provision built in. If not, oil will
be sucked from inside the vacuum pump back into the experiment.
To prevent this the experiment must be isolated from the pump by
closing a valve before the pump is switched off. Air should then be
allowed into the pump using the air inlet valve. The experiment can
be brought back to atmospheric pressure when needed using a sepa-
rate air inlet valve in the experiment: if the air inlet valve on the
pump is used oil will be sucked into the experiment.

2.5.5 DIFFUSION PUMP

In a diffusion pump molecules of gas are swept away by jets of

oil vapor. The exhaust must be connected to a rotary pump: that is, a
diffusion pump must be backed by a rotary pump. Diffusion pumps
have a high displacement rate even at low pressures and so will reach
low pressures despite minor leaks. They can achieve a pressure of
around 1×10−4 mbar.

2.5.6 TURBOMOLECULAR PUMP

A turbomolecular pump (see Figure 2.10) contains a rapidly

spinning turbine which repeatedly collides with gas molecules
imparting momentum, directing the molecules towards the exhaust.
The exhaust must be connected to a rotary pump: that is a turbomo-
lecular pump must be backed by a rotary pump and can achieve a
minimum pressure of around 1×10−8 mbar.
To prevent the turbine blades becoming damaged due to too
many collisions with gas molecules, turbomolecular pumps should
not be turned on until the rotary backing pump has achieved a pres-
sure of at least 9×10−3 mbar. Care must be taken when switching off
a turbomolecular pump. Air should not admitted to the system until
the turbo pump has slowed to 25–50 % of its maximum revolutions
24 • Physics Lab Experiments

Figure 2.10: A turbomolecular pump backed with a rotary pump.

 Safety • 25

per minute. This may take a considerable time of the order of

15–30 minutes. Air should then be admitted very slowly through the
special air inlet valve.

2.6 CRYOGENICS

Cryogenics refers to the production of very low temperatures in

the laboratory. Rather than using a refrigerator such as you might
find in a kitchen, in the physics lab liquid cryogens are generally
preferred. This is because of their simplicity, relatively low cost and
the complex nature of cryogen free coolers.
Common liquid cryogens are nitrogen and helium. These ele-
ments are gases at room temperature and pressure, however can be
cooled and liquified. Often cryogens are produced off site by a spe-
cialist company and purchased by a university. Liquid cryogens pro-
vide a simple reliable method to cool other objects to very low and
stable temperatures. It is likely that liquid nitrogen and dry ice will
feature early on in many undergraduate courses and so this section
begins by discussing how these can be used safely. Liquid helium
will also feature in some undergraduate laboratories and so this will
be mentioned briefly.
No written publication or instructions can never be a substitute
for proper training by a university before any cryogenic liquids are
used. This ensures that the student is aware of all the dangers specific
to the equipment and setting in which they will be using cryogens.

2.6.1 LIQUID NITROGEN

At atmospheric pressure nitrogen boils at 77 K (196°C) and it

has a latent heat of vaporization of 200 kJ/kg. In its liquid form it
is a clear, colorless liquid. Nitrogen freezes creating nitrogen ice at
63 K (210°C). By reducing the pressure of the gas above a container
of liquid nitrogen with a rotary vacuum pump it is possible to freeze
nitrogen. Figure 2.11 shows the phase diagram for Nitrogen.
26 • Physics Lab Experiments

Figure 2.11: Phase diagram for Nitrogen. Nitrogen is a gas at STP (standard temperature
and pressure).

Due to its high latent heat of vaporization and relatively low

cost it is commonly used to cool objects in the physics laboratory.
It will be stored in a Dewar flask (named for the Scottish scientist
James Dewar 1842–1923) which is essentially a large vacuum flask
of the type used for hot drinks: see Figure 2.13. If a vacuum flask
intended for food use is used it is vital the screw top is not used as
this will lead to a build up of pressure as the nitrogen evaporates
and a possible explosion. For easy access to small quantities it may
be temporarily stored in a polystyrene or neoprene bucket: see Fig-
ure 2.12.
It is reasonably safe to use in the laboratory provided there is
some understanding of the dangers it poses and simple safety proce-
dures are followed.
The primary danger is the cooling and freezing of parts of the
human body. However, if liquid nitrogen comes into gentle contact
 Safety • 27

Figure 2.12: A neoprene plastic bucket for storing liquid nitrogen.

with skin it rapidly evaporates forming a layer of nitrogen gas which

protects the body from any further contact with the liquid as well as
insulating the body from the coldness of the liquid. This is called the
Leidenfrost effect and can also be seen if water is spilt on a hot elec-
tric cooker plate. Thus it is perfectly safe to dip fingers into liquid
nitrogen for a short time (less than 1 second) while being careful to
avoid touching any container holding the nitrogen. It is also possible
to pour liquid nitrogen over hands or other parts of the body pro-
vided it is not poured from a significant height nor allowed to collect
in the cup of hands, pockets, sleeves, cuffs, or shoes. Liquid which
collects in clothing will be in contact with the body for a long period
of time and could lead to severe cold burns.
This possibility of collection of liquid absolutely rules out the
use of gloves when handling liquid nitrogen. Vitally however, solid
objects, especially metal, cooled by liquid nitrogen should never be
touched with bare hands: thick gloves (suede gardening gloves or
special cryogenic gloves may be appropriate) must always be worn.
Bare hands will stick to cold objects (rather than follow the recoil
28 • Physics Lab Experiments

Figure 2.13: Liquid Nitrogen Dewars. Left: a 25 liter “onion” Dewar with a cart Right: a 10
liter Dewar.

reaction experienced when hot objects are touched) and present the
danger severe cold burns.
Care should be taken to avoid splashes of liquid nitrogen enter-
ing the eyes especially when pouring liquid nitrogen from one con-
tainer to another and so wearing goggles may be appropriate. Open
shoes or sandals must be avoided by all those working in the same
room as liquid nitrogen is used.
A further danger presented by liquid nitrogen is asphyxiation.
When a liquid turns into a gas it expands significantly (remember
the diagrams showing the particles in a solid, liquid, gas drawn
 Safety • 29

at school). In the case of liquid nitrogen warming to room tem-

perature, the expansion factor is 683 (i.e., 1 liquid liter expands
to 683 gas liters at room temperature and pressure). Therefore
if even a small amount of liquid nitrogen is spilt in a small room,
lift, car or other confined space it can displace the oxygen causing
asphyxiation. This danger can be avoided by never traveling in a
car or a lift with liquid nitrogen and ensuring that the quantity
of liquid is always appropriate for the room in which it is being
stored or used.
This paragraph contains a detailed calculation regarding the
minimum room volume in which a 25 liter Dewar of liquid nitrogen
could safely be stored or used assuming there is no low oxygen alarm
linked to a fail safe forced ventilation system. The BCGA Code of
Practice [3] states that “From a normal oxygen level of 21%, a room
shall be sufficiently ventilated for the two, normal conditions not to
cause a reduction in oxygen concentration below 19.5%: (i) the nor-
mal evaporation of all dewars and liquid nitrogen containers within
the room (ii) the filling losses from filling the largest dewar from a
warm condition. Additionally, the complete spillage of the contents
of the largest dewar shall not cause the oxygen concentration to fall
below 18%”. Following Annex 3 of [3], accounting for the normal
evaporation rate of the 25 liter Dewar the concentration of oxygen
in the room after a long period will be
Vr  0.21  n
C (2.1)
L  (Vr  n)

where Vr is the room volume, L is the gas release units of m3/h and
n is the number of air changes in the room per hour. Assuming a
typical value of n=0.4 and that L is twice the manufacturers quoted
value of 0.2 liquid liters per day (i.e. L=0.4 × 683 = 273 gas liters
per day = 11.4 gas liters per hour = 0.0114 m3/h where the gas
expansion factor is 683) gives the minimum room volume to not
fall below 19.5 % oxygen as 0.38 m3. This is insignificant compared
to the requirement concerning the complete spillage of a Dewar.
Following Annex 4 of the same code, accounting for the complete
spillage of the 25 liter dewar the minimum room volume for stor-
age/use will be given by
30 • Physics Lab Experiments

 Vd  f g 
0.21  Vr –
 1000  (2.2)
C
Vr

where C is the oxygen concentration, Vd is the dewar capacity in

liters and fg is the gas expansion factor. Solving this leads to a mini-
mum room volume of 120 m3 that the oxygen concentration will not
fall below 18 %. This should be appropriately adjusted if more than
1 Dewar is present.
Where asphyxia is possible (i.e., where liquid nitrogen has been
used or spilled in a confined space) if any of the following symptoms
appear immediately remove the affected person to the open air, and
follow up with artificial respiration if necessary: rapid and gasping
breathing, rapid fatigue, nausea, vomiting, collapse, or incapacity to
move or unusual behavior.
First aid for cold burns should include running the affected
area under warm (but not hot water), covering with clean cling film
and a prompt visit to the accident and emergency department at a
hospital.

2.6.2 DRY ICE

Dry ice is the name given to solid (frozen) carbon dioxide

because it sublimes (i.e., turns directly from a solid to a gas with no
liquid phase) as its temperature increases above 194.5 K (78.5°C).
This can be seen by studying the phase diagram in Figure 2.14. Its
density when solid at 194.5 K and 1 atmosphere is 1560 kg/m3 and
when a gas at 273 K and 1 atmosphere is 1.98 kg/m3. This is more
dense than air (1.3 kg/m3) so it sinks to the bottom of a container or
a room and any leaks large releases of gas will fill the room from the
floor upwards.
Dry ice is usually supplied in the form of white cylindrical pel-
lets around 8 mm in diameter and 1–2 cm in length. It can usu-
ally stored in a thick polystyrene box with a loosely fitting lid for a
few days.
 Safety • 31

The primary danger from handling dry ice is severe frost bite,
therefore dry ice should be handled only while wearing thick gloves
(suede gardening gloves or special cryogenic gloves may be appro-
priate). Similar gloves should also be used while handing objects
cooled by dry ice. With extreme care, individual pieces of dry ice my
be handled without gloves provided they are not gripped tightly nor
held for more that a second or two.
As with liquid nitrogen, care must be taken over storage espe-
cially in small room or confined spaces. The typical carbon dioxide
content of the atmosphere is around 390 ppm (parts per million) [4].
The safe limit is considered 5000 ppm, above this level symptoms
consist of an increased breathing rate, tiredness, and headaches. If
any of the symptoms appear immediately remove the affected person
to the open air and follow up with artificial respiration if necessary.

Figure 2.14: Phase diagram for Carbon Dioxide. Carbon Dioxide is a gas at STP
(standard temperature and pressure).
32 • Physics Lab Experiments

2.6.3 LIQUID HELIUM

At atmospheric pressure liquid helium boils at 4.2 K (269°C)

and it has a latent heat of vaporization of 21 kJ/kg. Helium does not
solidify at atmospheric pressure. As a gas it is less dense than air with
a density of 0.179 kg/m3 at 273 K and 1 atmosphere meaning that
any leaks or large releases of gas will fill the room from the top down.
Figure 2.15 shows the phase diagram for Helium.
Submerging objects in liquid helium allows their temperature to
be reduced much further than when liquid nitrogen is used, however
it is significantly more expensive to produce than liquid nitrogen.
Universities with large low temperature physics groups may have a
helium liquefier, although often liquid helium is purchased from an
external company. If a liquefier is present, experiments will usually

Figure 2.15: Phase diagram for Helium-4. Helium-4 is a gas at STP

(standard temperature and pressure).
 Safety • 33

be connected to a recovery line so that any liquid helium which boils

off from the experiment is returned to be re-liquefied.
Safety precautions are similar to those for liquid nitrogen. How-
ever, before filling a container or vessel with liquid helium, the air in
the container should be removed using a rotary vacuum pump and
then replaced with helium gas from a cylinder. This prevents the
liquid helium solidifying any nitrogen (or small amounts of oxygen,
carbon dioxide, or water vapor) which may be present, leading to a
blockage in the equipment.
If objects need to be cooled to 4.2 K they would usually be pre-
cooled with liquid nitrogen due to the expense of liquid helium and
the significantly higher latent heat of vaporization of nitrogen. This
means that only about a tenth of the mass of liquid nitrogen com-
pared to liquid helium is needed to cool an object from room tem-
perature to 77 K.

2.6.4 HELIUM-3

Helium-3 or 32He is a isotope of helium which has a nucleus

consisting of 1 neutron (instead of the more usual 2 neutrons) along
with 2 protons. At atmospheric pressure Helium-3 boils at 3.2 K
(-270°C) and can’t be solidified. Helium-3 is found naturally along
with Helium-4, although it is approximately 10,000 time less com-
mon than Helium-4. Obtaining Helium-3 for use the in laboratory is
incredibly expensive (approximately a few tens of thousands of GBP
per gram in 2012) and so extreme care is taken when it is used.
The reason for its use is that it enables temperatures below
around 2 K (271°C) to be consistently and reliably achieved. If a
Helium-4 bath has a large rotary vacuum pump attached to it the
pressure of the vapor above the liquid can be reduced. This leads
to the Helium-4 liquid cooling down to around 2 K. This is cold
enough to liquify a separate jacket of Helium-3 around the sample.
If a second vacuum pump is attached to the Helium-3 jacket and
the pressure reduced, the liquid Helium-3 will reduce in tempera-
ture to around 0.65 K. Due to the very high cost of Helium-3, great
care must be taken when using it. It will have its own dedicated
34 • Physics Lab Experiments

pump with a closed exhaust and a reservoir. When using a Helium-3

system if at all unsure what to do or what opening/closing a particu-
lar valve does always ask for help. This is much better than losing or
contaminating the Helium-3.
When the pressure is reduced above a gas, the pressure can
be used as a guide to the temperature of the liquid. Tables and
give the temperature of liquid Helium-4 and Helium-3 at differ-
ent pressures [5].
A more advanced use of Helium-3 and Helium-4 mixtures is
in a dilution refrigerator. These can achieve temperatures of the
order of a few milliKelvin but a full explanation is beyond the
scope of this book.

Table 2.5: Table of temperature and vapor pressures for He-4

 Safety • 35

Table 2.6: Table of temperature and vapor pressures for He-3

2.6.5 LIQUID OXYGEN

At atmospheric pressure liquid oxygen boils at 90 K (183°C). In

its liquid form it is a translucent, pale blue liquid. Liquid oxygen is also
paramagnetic, meaning that it develops an attraction to an externally
applied magnetic field. Oxygen freezes creating oxygen ice at 50 K
(223°C). Figure 2.16 shows the phase diagram for Oxygen.
From the point of view of liquid oxygen being a cryogen, safety
procedures are similar to those for liquid nitrogen. It is frequently
thought that liquid oxygen is highly flammable and therefore very
dangerous. This is misleading: liquid oxygen is an accelerant. This
means that it encourages fuels to burn more quickly and intensely
than they would if liquid oxygen was not present. If fuels and sources
of ignition are kept away from liquid oxygen there should be no sig-
nificant safety concerns in making or using small quantities.
The usual way to liquify oxygen in the laboratory is to pass oxygen
gas from a compressed gas cylinder through a coil made from hol-
36 • Physics Lab Experiments

Figure 2.16: Phase diagram for Oxygen. Oxygen is a gas at STP (standard temperature and
pressure).

low copper pipe which is submerged in liquid nitrogen. The copper

coil is a good conductor of heat and has a large surface area. Liquid
oxygen is then collected in a thermos flask or a small dewar. How-
ever, the lack of a cylinder of compressed gas does not prevent liquid
oxygen being made [6].

2.6.6 ARGON

Argon gas is sometimes used to provide an inert atmosphere

in which reactive elements or compounds can be kept for an
extended period of time. At atmospheric pressure liquid argon
boils at 87 K (186°C) meaning that it can be liquefied in the labo-
ratory using liquid nitrogen in a similar way to the method used
 Safety • 37

to produce liquid oxygen. Argon is colorless and odorless in gas,

liquid and solid forms. Argon freezes creating argon ice at 84 K
(189°C).
As a gas it has a density of 1.78kg/m3 which is greater than the
density of air. This means is sinks to the bottom of a container or a
room and any leaks will fill the room from the floor upwards. Figure
2.17 shows the phase diagram for Argon.

2.6.7 CRYOGEN FREE COOLING

Currently cryogen free or closed cycle cryogen systems are

expensive, so it is reasonably unlikely that these will met in the
undergraduate physics laboratory. However, over time as prices

Figure 2.17: Phase diagram for Argon. Argon is a gas at STP

(standard temperature and pressure).
38 • Physics Lab Experiments

fall and as liquid cryogens become increasingly expensive they will

become more common.
Closed cycle cryogen systems aren’t cryogen free. They are
closed systems so that as the cryogen is boiled into a gas, the gas is
collected and then recondensed using an electrically powered refrig-
eration system. They are sometimes classified as cryogen free as the
cryogen does not need replacement.
True cryogen free systems operate on a broadly similar principle
as a domestic fridge/freezer typically with helium as the working
gas. Two types which are common are pulse tube refrigerators and
Gifford-McMahon (GM) coolers.

2.7 LASERS

Lasers can be immensely useful in the undergraduate

physics laboratory in a wide range of experiments as they provide
monochromatic (single wavelength), coherent (constant phase
relationship) light. Laser is an acronym for Light Amplification by
Stimulated Emission of Radiation.
Although some lasers are perfectly safe, others will cause
almost immediate retina damage and blindness if they are viewed
even indirectly. Lasers are classified into four classes (and a few
subclasses) based on their wavelength and maximum power out-
put. Recently, a new classification system has been adopted,
although lasers may still be seen labeled with the old classification
system.
The new classification system is shown in Table 2.7.
Class 1: A class 1 laser is safe under all conditions
of normal use and should not lead to eye
damage even if magnified using a telescope or
microscope. Devices should be fitted with a
warning stating “Class 1 laser product”.
 Safety • 39

Class 1M: A class 1M laser is safe under all condition

of normal use except magnification with
a telescope or microscope. The laser can
move to a higher class if the beam is focused
rather than having a large diameter or being
divergent. Devices should be fitted with a
warning stating “Laser radiation do not view
directly with optical instruments. Class 1M
laser product”.
Class 2: A class 2 laser is safe because the typical
human blink reflex limits the exposure. Only
visible lasers (400–700 nm) can be class 2 and
they are limited to 1 mW for a continuous
beam. Deliberate avoidance of blinking can
lead to eye damage. Many laser pointers and
measuring instruments are class 2. Devices
should be fitted with a warning stating “Laser
radiation do not stare into beam. Class 2 laser
product”.
Class 2M: A class 2M laser is safe because of the blink
reflex unless it is viewed through optical
instruments such as telescopes or microscopes.
Devices should be fitted with a warning stating
“Laser radiation do not stare into beam or view
directly with optical instruments. Class 2M
laser product”.
Class 3R: A class 3R laser is generally considered
safe if care is taken to restrict viewing of the
beam. Visible lasers are limited to 5 mW for a
continuous beam. Devices should be fitted with
a warning stating “Laser radiation avoid direct
eye exposure. Class 3R laser product.”.
40 • Physics Lab Experiments

Class 3B: A class 3B laser will cause damage to the eye

if directly exposed although diffuse reflections
(from matt surfaces such as walls or paper) are
not harmful. Typically, visible wavelength lasers
are limited to 30 mW. Protective goggles are
required for the eyes where there is a possibility
of direct viewing. Goggles must absorb the
wavelength of light emitted by the laser and will
typically be colored. Class 3B lasers are used in
CD and DVD writers, however the drive unit
is Class 1 because the laser light cannot escape.
Devices should be fitted with a warning stating
“Laser radiation avoid exposure to beam. Class
3B laser product”.
Class 4: Class 4 lasers are the most dangerous. They
can cause permanent eye damage, burn skin
and ignite combustible material from direct
viewing and even from diffuse reflections.
Great care should be taken to control the path
of the beam. Many lasers in use in scientific
research, medical uses and industry fall into
this category. Devices should be fitted with a
warning stating “Laser radiation avoid eye of
skin exposure to direct or scattered radiation.
Class 4 laser product”.
Table 2.7: The new laser classifications.

The older classification system which has now been superseded may
still be in use on older devices. This is shown in Table 2.8.

2.8 IONIZING RADIATION

Ionizing radiation is given off by radioactive isotopes in the

form of alpha and beta particles and gamma rays. There may also be
 Safety • 41

positrons, X-rays, and further conversion electrons emitted. Radio-

active particles are present naturally in the surroundings and con-
tribute to a background count of around 3–4 counts per 10 seconds
if measured with a Geiger counter.
There are a range of units in which radioactivity are measured,
each with slightly different meanings. The most straightforward is
the Becquerel: 1 Bq corresponds to 1 nucleus disintegrating per
second (note that a single nucleus may give rise to the emission
of more than one radioactive particle). An alternative unit is the
Curie: 1 Ci corresponds to 37 GBq. Knowing the number of dis-
integrations doesn’t give any information about the absorbed dose
a person may receive. This is measured in terms of the energy
deposited per kilogram or Greys: 1 Jkg−1 = 1 Gy. However the
harmfulness of different types of radioactive particles varies so
there is a weighting factor which is applied to the absorbed dose
in Gy to give an equivalent dose in Sieverts (Sv). This weighting
factor is 1 for photons, electrons, and muons. It is 5 for protons >
2 MeV, and neutrons < 19 keV and > 20 MeV. It is 10 for neutrons
in the ranges 19 keV to 100 keV and 2 MeV to 20 MeV. It is 20
for neutrons in range 100 keV to 2 MeV and for alpha particles,
fissions fragments and heavy nuclei. Finally an effective dose (also
measured in Sv) is found by multiplying the equivalent dose by
a tissue weighting factor accounting for the varying sensitivity
of different areas of the body to radiation. A typical background
radiation count over a year is 3 µSv. In the UK, the Ionizing
Radiation Regulations 1999 set dose limits for the general public
of 1 mSv/year, for non-classified workers of 6 mSv/year and for
classified workers of 20 mSv/year.
If sources are used with thought and care there is little chance
of a dose getting any where near as high as 3 µSv which corresponds
to an extra 1 year of background radiation or the dose received in a
single dental X-ray. Ionizing radiation can be used by anyone over
16 provided they are responsible and have had training in the safe
use of the source.
Sources will commonly be sealed in a durable cup holder and
kept in a lead lined box: see Figure 2.18. Sources should always be
kept secure when not in immediate use. The university have a legal
42 • Physics Lab Experiments

Class I: Similar to Class 1.

Class II: Similar to Class 2.
Class IIa: Similar to Class 2, although lower power.
Direct viewing for up to 1000 seconds is
considered safe.
Class IIIa: Similar to Class 2M, although direct viewing
for more than 2 minutes will cause retina
damage.
Class IIIb: Similar to Class 3B.
Class IV: Similar to Class 4.
Table 2.8: The old laser classifications.

Figure 2.18: Left panel: a lead lined wooden storage box and forceps. Right panel: inside
the box and the source inside a metal cup holder.

duty to know the exact whereabouts of every source they own:

in practice this means that sources are signed out of a locked
storage cabinet and then signed back in when they are no longer
in use. Sources should be carried and kept as far away from the
body as reasonably practical and only handled with long forceps.
Be aware of leaving unused gamma sources within a few meters
of an experiment as this radiation may erroneously be measured.
Latex/nitrile gloves should be worn when using any sources
which are not sealed in cup holders. This prevents any flakes of
radioactive material becoming trapped in clothes or under finger
nails.
 Safety • 43

Table 2.9: List of common radioactive sources. Protactinium and Radon are listed with
radiation given out by the decay chain which generates them from Uranium-238 and Tho-
rium-232 respectively.

Alpha particles have a range in air of up to a few cm. Beta par-

ticles have a significantly further range in air (around 1 m), but can
be blocked by the lead lining in the source box or a few millimeters
of aluminium. Gamma rays are more penetrating and will not be at
all collimated. They will emit equally in all directions from the cup
holder. However, gamma rays follow the inverse square law, so if
the distance between the source and the user is doubled, the dose
received will be quartered.
Detection of ionizing radiation is usually by a Geiger-Muller
tube. This is a cylinder with an axial electrode filled with low pres-
sure gas. A potential difference of a few hundred volts is applied
between the cylinder and the axial electrode. When a beta particle
enters through the mica window at the front it will ionize one of
the gas molecules creating a positive ion and an electron. These
particles will be accelerated to opposite electrodes, when they gain
sufficient kinetic energy they will be able to cause secondary ioniza-
tion events and then an avalanche effect which produces a measur-
able current which can be registered by a counter. Gamma particles
will cause the same process, but do not need to enter the tube at
the end through the mica window as they will easily penetrate the
tube from any direction. Finally, alpha particles are only detected
from an extremely close range and even then many are blocked
44 • Physics Lab Experiments

by the mica window. A spark counter is a more appropriate way to

detect alpha particles. This is a pair of wires across which a very high
(electric) potential difference is applied. When an alpha source is
brought near, the resulting ions created in the air causes a spark to
jump between the wires.
Common sources are shown in Table 2.9 along with their halflife
and main radiation(s).
 CHAPTER

3
MAKING
MEASUREMENTS

When using a new piece of apparatus, try to perform a pre-

liminary experiment in which a basic measurement is made. This
will increase familiarity with the equipment and confidence that
it is giving a reliable result. For example, with a multimeter, the
(electric) potential difference of an AA battery could be measured,
expecting the outcome of around 1.5 V. Alteratively measure the
resistance of a known resistor. A preliminary experiment should
also allow a check to be made that the equipment is working cor-
rectly, that it has a suitable range and precision for the experiment
and allow an estimate to be made of the uncertainty or error in the
measurements.
This section focuses on a range of measuring devices which
may be encountered, explaining how they are used and some-
times the underlying physics. Care should always be taken when
reading analogue or pointer style meters and scales to avoid par-
allax errors. These arise when the eyes are not directly above the
scale meaning the pointer may line up with an incorrect mark on
the scale. Taking large numbers of decimal places from digital
meters should also be avoided unless it is known the meter has
been calibrated to this resolution and that the result is genuinely
meaningful.
46 • Physics Lab Experiments

3.1 SI UNITS

The International System of Units is abbreviated SI from the

French, Le Système International d’unités. It gives a set of seven
base units in which all units can be expressed: see Table 3.1. For
example the Newton is equivalent to the kgms−2 and the Joule is
equivalent to kgm2s−2.

Table 3.1: The SI units

Table 3.2: The SI prefixes

 Making Measurements • 47

3.2 SI PREFIXES

Prefixes are used to show multiples and submultiples of SI units.

Table 3.2 shows a list of values.

Figure 3.1: A Vernier Scale. The top is the main scale and the bottom in the traveling scale.
The left panel shows a reading of 0.0 mm; The middle panel shows a reading of 0.1 mm;
The right panel shows a reading of 0.6 mm.

3.3 VERNIER CALIPERS

Vernier calipers (usually with a resolution of 0.01 mm) are used

to measure very small distances, such as the thickness of a wire, pre-
cisely. Achieving an accurate reading with Vernier calipers is depen-
dent on the skill of the operator. The object must always be clamped
squarely between the jaws of the calipers which must be forced
into contact with the object to be measured. As both the object
and the calipers are slightly elastic, if too much force is applied and
the object squashed the calipers will under-read. Conversely, if not
enough force is used the calipers will over-read.
A Vernier scale is two adjoining traveling scales with slightly
different spacings. The main scale, for example, has divisions every
1 mm. The lower scale has divisions every 0.9 mm. So 10 divisions
on the lower scale take 9 mm. This means the tenth division on the
lower scale lines up exactly with 9 divisions on the main scale. This
can be seen in the left panel of Figure 3.1.
As the lower scale is moved by 0.1 mm, the first division on the
lower scale will line up with the first division on the main scale. This is
shown in the middle panel of Figure 3.1. The right panel of Figure 3.1
shows a reading of 0.6 mm.
48 • Physics Lab Experiments

Figure 3.2: A digital calipers measuring; Left: the diameter of a wire and Right: the inside
diameter of a pipe.

For objects larger than 1 mm, the number of millimeters on the

main scale immediately to the left of the zero mark on the traveling
scale must be added to the reading from the traveling scale.

3.4 DIGITAL CALIPERS

Digital calipers are becoming more and more common as a

replacement for Vernier calipers as they are easier and quicker to
use. The jaws are closed and the reading is zeroed, usually by press-
ing the “zero” button. The object to be measured can then be held
in the jaws and the distance read from the display. Figure 3.2 shows
a digital calipers being used to measure a thin wire (0.68 mm) and
the inside diameter of a cylindrical pipe (13.95 mm).

3.5 MICROMETER

Micrometers use a screw to convert a small linear distance

moved into a large rotation which can be read from a scale. A Ver-
nier style scale is also sometimes incorporated to increase the pre-
cision of the micrometer. The screw is usually fitted with a ratchet
mechanism which allows objects to clamped with a consistent and
 Making Measurements • 49

Figure 3.3: A micrometer measuring a: 5.21 mm and b: 3.76 mm

appropriate force which reduces error due to the elastic nature of

the object and the micrometer.
To make a measurement with a micrometer an object is posi-
tioned between the jaws. The ratchet screw is turned until it clicks
indicating the object is correctly pressed between the jaws. A read-
ing of the lowest half millimeter is taken from the fixed stem of the
micrometer. The alignment of the rotating barrel of the micrometer
with the axial line on the stem provides an extra decimal place to the
reading. For example Figure 3.3a shows the micrometer displaying
5.21 mm and Figure 3.3b shows the micrometer displaying 3.76 mm.
For Figure 3.3a:
Stem Reading 5.00 mm
Barrel Reading 0.21 mm
Total 5.21 mm
For Figure 3.3b:

Stem Reading 3.50 mm

Barrel Reading 0.26 mm
Total 3.76 mm

3.6 BALANCES

Digital balances provide an indirect method for finding the

mass of an object. Balances which measure to many mass ranges and
50 • Physics Lab Experiments

precisions are available and it should always be ensured that the bal-
ance used is appropriate for the measurement needed. For example:
it would not be appropriate to use a 0–2 kg balance which reads
to the nearest 1 g to try to measure 100 mg of metal, neither is it
appropriate to check the mass of a 1 kg standard mass with a 0 - 500
g balance which reads to the nearest 1mg.
Balances involve the use of one or more pressure sensors which
produce an electrical output dependent upon the force acting on
them. This electrical output has been calibrated by the manufac-
turer as corresponding to a particular force. This force is then con-
verted to a mass by dividing by an average value for the acceleration
due to gravity. According to the latest data from the NASA Grace
mission the acceleration due to gravity varies around the Earth by at
least 0.001 ms−2. Thus quoting masses to more than 4 significant fig-
ures is inappropriate on balances not calibrated for use in a specific
location, although this may not be true for mass differences.
The balance should be positioned directly on a flat surface: ensure
there are no wires or other objects under the balance. When mak-
ing measurements on a table, desk or worksurface ensure that no one
touches, leans or sits on the surface as this can sometimes significantly
alter the reading on the scale. It is useful to perform a quick experi-
ment to see the significance of this. Some balances have a cover to
prevent draughts (possibly from people walking past or convection
currents from heating or air conditioning) affecting the results: this
should be placed down and closed when any readings are taken.
To take a measurement first ensure that the scale pan is clean and
free from loose material and the balance reads zero. If not the “Tare”
or “Zero” button should be pressed. If a container is used often to
hold powder which is being measured out, this can be placed on the
balance before it is zeroed. Finally add the object to be measured to
the scale pan and take a reading.

3.7 AMMETERS

Ammeters measure the current flowing in a circuit. Current is

the rate of flow of charge. They are placed in series with the com-
ponent through which the current is to be measured. To cause the
 Making Measurements • 51

minimum (or no) change to the behavior of the rest of the circuit an ideal
ammeter will allow current to flow through with no resistance. In prac-
tice ammeters have a very low resistance, which is assumed to be small
compared to the circuit being measured. To picture this more clearly
consider the equation for the total resistance of resistors in series:

RT  R  Rammeter (3.1)

It is required that RT is as close as possible to the resistance, R of

the component. This can be achieved by making Rammeter as small as
possible.

3.8 VOLTMETERS

Voltmeters measure the (electric) potential difference (some-

times called the voltage drop) across a component or between two
points in a circuit: this is the difference in energy per coulomb of
charge. They are placed in parallel with the component across which
the (electric) potential difference is to be measured. To cause the

Table 3.3: Resistor color codes

52 • Physics Lab Experiments

minimum (or no) change to the behavior of the rest of the circuit an
ideal voltmeter will have an infinite resistance and allow no current
to flow through it. In practice, voltmeters have a very high resis-
tance, which is assumed to be large compared to the circuit being
measured. To picture this more clearly consider the equation for the
total resistance of resistors in parallel:

1 1 1
  (3.2)
RT R Rvoltmeter

It is required that RT is as close as possible to the resistance, R of

the component. This can be achieved by making Rvoltmeter as large as
possible.

3.9 RESISTOR COLOR CODES

Resistors typically have 4 colored lines indicating the size of the

resistance. Table 3.3 indicates how to find the resistance.

3.10 MULTIMETERS

Frequently ammeters and voltmeters used in the lab will

actually be part of a multimeter device. Multimeters are versatile
devices capable of making measurements of variables such as cur-
rent, (electric) potential difference in both AC and DC circuits and
resistance. Figure 3.4 shows two typical multimeters. The round
dial in the center selects which variable the meter measures as well
as the maximum reading measurable in that range. The  symbol
indicates a resistance measurement, the A symbol indicates an
ammeter and the V symbol indicates a voltmeter. The symbol of a
solid line above a dashed line indicates DC and the sin wave symbol
indicates AC. The setting 20 µA indicates that the maximum read-
ing measurable will be 20 µA and the reading given will be in units
of µA. The setting 200 k indicates that the maximum reading mea-
surable will be 200 k and the reading will be given in units of k.
 Making Measurements • 53

Figure 3.4: Two different multimeters.

Two wires must be connected to the front of the multimeter: one

to the black common point and the other to a red point correspond-
ing with the variable being measured.
To measure a resistance using only two connections the
multimeter generates a known constant current which it passes
through the component or circuit connected to the meter. It then
acts as a voltmeter and measures the potential difference across
the component. The simple calculation of R = V/I gives the resis-
tance.

3.11 FURTHER IDEAS ON MEASUREMENT OF

RESISTANCE

The usual method of attaching two wires to a component/

sample of material to measure its resistance involves passing cur-
rent and measuring the (electric) potential difference through the
same wires. This has the disadvantage of measuring the sum of the
component/sample resistance, contact resistance and the resistance
54 • Physics Lab Experiments

Figure 3.5: Simple diagram of a four-wire method of measuring the sample resistance. R1,
R2, R3, and R4 are resistances which represent the contact and lead resistances.

of the measurement wires. In some cases where the component/

sample resistance is much higher than the lead resistance this is
acceptable as the additional resistance only forms a very small part
of the measurement. In the case of samples with resistance O(m)
an alternative method must be found especially if the measurement
wires have a high resistance (perhaps due to their narrow diameter
to reduce the heat load on a cryogenic experiment).
Resistance can be measured by using the four wire method
(Figure 3.5) where four wires are separately attached to the sam-
ple. Two are used to pass current through the sample and a further
two are used to measure the (electric) potential difference across
the sample. The source current flows through R1, the sample and
R2 and because no current flows through R3 and R4 the (electric)
potential difference measured by the voltmeter is just that across
the sample.
 Making Measurements • 55

Figure 3.6: Simple four-wire mounting configuration of a sample to measure resistance

along in the direction of the current flow.

Figure 3.6 shows a contact configuration which can be used to

measure the resistance of a sample of conductive material. The cur-
rent is passed through the sample via the two surfaces at each end
and the (electric) potential difference is measured over a distance l
between the other two contacts. The voltage contacts should be far
away from the current contacts to ensure the current flow is uniform
and parallel between the voltage contacts. The current contacts
should cover the whole of the end surfaces to ensure a uniform flow.

3.12 CONSTANT CURRENT SOURCES

When making electrical measurements in which the resistance

of the sample might change, a constant current source is needed.
This is not trivial to construct as a cell, battery, or normal power sup-
ply all have internal resistance. Recalling that:

I
R r (3.3)

where I is the current,  is the EMF of the power source, R is the

external resistance and r is the internal resistance of the power
source. If R changes, but  and r remain constant then the current
in the circuit changes. R and I have a non linear relationship which
makes calculation of R given V and I very difficult.
56 • Physics Lab Experiments

An ideal current source should output the same current regard-

less of the resistance of the sample or the leads. A number of
different methods of generating constant current can be used.
The most straight forward method is to use a variable resis-
tor Rseries in series with the sample. An AC or DC supply can be
connected to this. Alternatively, if a lock-in amplifier is used they
have an internal oscillator which outputs an AC EMF in the range
0 – 5 V. Typically resistors between Rseries=1 k and 10 k can be
connected to the lock-in amplifier front panel oscillator output and
to the sample as shown on the left in Figure 3.7. It is easily seen that
the current flowing through the sample is given by:

V V
I  (3.4)
R Rseries  Rsample

Rseries >> Rsample otherwise as the sample resistance changes for

any reason (perhaps due to a change of temperature) the current
going through the sample will also change leading to ambiguous
results. The value of Rseries used is limited by the fact that most
lock-in amplifiers can only output 5 V; when a larger R is used, the
current becomes very small. Currents need to be sufficiently large
to produce a high quality signal above any background noise as well
as sufficiently low so that no self heating is detectable in measure-
ments.
An alternative method is to use the (electric) potential differ-
ence controlled current source circuit as shown on the right in
Figure 3.7. For an op-amp with the non-inverting(+) input con-
nected to ground, the (electric) potential difference between the
inverting(-) input and ground is also zero. An ideal op-amp has an
infinite input impedance so draws no current from its input circuit
implying the sum of the currents (Kirchoff’s First Law) at point A
should be zero:

Vin V
 – B (3.5)
Ri Rf

Where Vin is the (electric) potential difference from the lock-in oscil-
lator output; Ri and Rf are the values of the resistors and VB is the
 Making Measurements • 57

Figure 3.7: Left panel shows a circuit layout of a resistor (typically Rseries = 1 kΩ to 10 kΩ)
used as a constant current source. It is connected in series with the sample and the oscillator
output on a lock-in amplifier. Right panel shows a circuit diagram of a (electric) potential
difference controlled constant current source. The op-amp used is an Linear Technology
LTC1150CN8#PBF (RS order number 5455629).

(electric) potential difference between point B and ground. Sum-

ming the currents going through the sample gives:

VB V
Isample   B (3.6)
Rg Rf

Combining Equations 3.5 and 3.6 gives Isample which is independent

of the sample resistance and is proportional to Vin.

R f  Rg
Isample  – V (3.7)
RiRg in

The lock-in oscillator output is connected to Vin and the sample cur-
rent contacts are those attached to the op-amp output. The current
in the sample is given by Equation 3.7.
A final method of generating a constant current is to use a dedi-
cated off the shelf constant current source such as a Keithley 6221
AC and DC current source. This has two current output leads which
are connected to the sample.
58 • Physics Lab Experiments

3.13 LOCK-IN AMPLIFERS

At a very simplistic level this can be thought of as a voltmeter. It

accurately measures very small AC signals by using phase-sensitive
detection to extract the signal from an often huge background that
would otherwise obscure the signal.
The sample is excited by an AC current at fixed frequency, wr, and
the lock-in detects the response of the sample at the same frequency
along with any background noise, 
V f ,sig Sin(w f t  qf ,sig ). It uses a
f
phase sensitive multiplier to multiply this by a self generated refer-
ence signal at the same frequency as the current VL sin(wrt  qref ).

Vpsdoutput  V
f
f ,sig VLsin w f t  q f ,sig  sin wr t  qref 
(3.8)

Vpsdoutput  2V
1
sig 
VL cos wr – w f  t  qsig – qref 
f
(3.9)
–
1

V V cos wr  w f  t  qsig – qref
2 sig L 
For each component frequency, f, that is detected this output is two
AC signals, one at the difference frequency wr−wf and the other at the
sum frequency wr+wf, which are then passed through a low pass filter.
In the general case for wr and wf there is no signal from the low pass
filter as both signals are AC and are filtered out. If wr=wf then part of
the signal is no longer AC, but DC and proportional to the signal ampli-
tude Vlpfoutput  2 Vsig VL cos qsig – qref  . This gives the X output of the
1

lock-in. The Y output is obtained by using a second phase sensitive

detector to multiply the detected signal by the reference frequency
shifted by 90° which gives an output Vlpfoutput  1 Vsig VL sin qsig – qref  .
2
Changing the offset of the reference frequency, qref, alters “the
phase” (qsig − qref) of the measurement. For a pure resistor in an AC
circuit the voltage response is exactly in phase with the current so
the phase should be set to zero (qsig − qref = 0) meaning that the X (“in
phase”) output of the lock-in gives the measured signal (cos(0)=1)
 Making Measurements • 59

and the Y (“out of phase”) output is zero (sin(0)=0). For an inductor

the current lags the voltage by 90° and for a capacitor the current
lead the voltage by 90° thus both these will show up on the Y (“out
of phase”) output.
The frequency of the excitation current, wr, can be altered on
the lock-in. Typically it would be in the region 50–100 Hz: but taking
care to avoid the mains electricity frequency (50 Hz in UK and 60
Hz in USA) and its harmonics (factors or multiples).
Figure 3.8 shows the front panel of a Stanford Research Sys-
tems (SRS) 830 lock-in amplifier. Usually BNC cables would be con-
nected to the “A” and “B” inputs in the bottom left of the front panel
and to the “SIN OUT” or “REF IN” BNC connectors at the bottom
right of the front panel.
First, a frequency must be selected for the lockin amplifier to
use. The right hand display shows the frequency when the “Freq”
button is pressed. It can be adjusted by rotating the dial on the right
of the front panel. Often the internal frequency generator is used
which outputs a sine wave via the “SIN OUT” connector. This is
then used within the circuit - for example see Section 3.12 on con-
stant current sources. If the internal frequency generator is used,
the amplitude can be displayed by pressing the “Amp” button and
can be changed by rotating the dial.
The frequency can also be selected by supplying a sine wave
or squarewave to the “REF IN” input. The “Source” button can be
pressed to switch between the “REF IN” signal and the internal fre-
quency. When the lock-in amplifier has not identified a signal to lock
on to the “UNLOCK” light is illuminated.

Figure 3.8: The front panel of a Stanford Research Systems (SRS) 830 lock-in amplifier.
60 • Physics Lab Experiments

The left display (channel 1) would usually be set to show the

“X” input which is the in phase component of the signal. The right
display (channel 2) would then be set to show the “Y” input which is
the out of phase component of the signal. The parameter displayed
can be changed by pressing the “Display” button.
The sensitivity panel displays the maximum (electric) potential
difference which can be measured in the present range. The range
can be changed by using the arrow keys to cycle through the avail-
able sensitivities/ranges. Underneath the X and Y display there is
a red “bar graph”. This gives a visual representation of size of the
signal in proportion to the present range. The range does not change
automatically and so needs to be adjusted as the signal amplitude
changes.
The time constant setting gives the characteristic time period
over which the outputs are averaged. A long time constant can be
used to smooth out very noisy signals, but this will increase the
response time.
To measure the true signal from a sample the phase should
be set correctly. The right hand display shows the phase when the
“Phase” button is pressed. For a resistance measurement, as a
first approximation, assume the sample is purely resistive and the
experimental equipment is sufficiently well designed that it doesn’t
cause large capacitive or inductive signals. The phase of the mea-
surement can be set using the “Phase” button on the “Auto” panel
of the lock-in. This sets the phase such that the Y output is zero.
In most cases this is a perfectly adequate assumption to make. In
some cases such as when the sample contacts are not of optimal
quality the phase of the measurement needs to be set to account
for the phase shift of the true resistive component of the signal.
Firstly, the normal measurement circuit shown in the left panel of
Figure 3.9 is modified to include a series resistor, approximately
of the same resistance as the sample giving the circuit in the right
panel of Figure 3.9. The (electric) potential difference across this
resistor is then measured with the lock-in. In the ideal case this is
a pure resistor so the auto-phase button is pressed to set the phase
such that the out-of-phase component, Y, is zero, the phase is then
kept at this value when the circuit is then returned to the original
configuration for the experiment.
 Making Measurements • 61

Figure 3.9: Left panel shows the circuit setup for a normal 4-wire measurement configura-
tion. Right panel shows the circuit setup when setting the phase. Current is passed though
the sample and a series resistor, R. The (electric) potential difference drop is measured
across only the resistor.

3.14 FURTHER IDEAS ON REDUCING NOISE IN

ELECTRICAL MEASUREMENTS

There are a number of possible sources of noise/error which

might affect electrical measurements such as external electromag-
netic fields, ground loops, capacitive coupling, self-inductive effects
and inductive cross-talk between wires.
The total impedance of a circuit is given by the sum

Z X2  Y2 (3.10)

 R 2  XC2  X L2 (3.11)
 
 R 
    L (3.12)

where R is the resistance, XC is the reactance due to the capacitive

voltage component with a phase which lags the current by /2 and
XL is the reactance due to the inductive voltage component which
leads the current by /2. This means that the Y output on the
lock-in amplifier contains all the (electric) potential differences of
capacitive and inductive origin and is dependent on the frequency of
the excitation current wr and the X output contains just the resistive
component and is independent of frequency. Clearly the resistive,
capacitive and inductive components can also be temperature
dependent.
62 • Physics Lab Experiments

The effect of external fields can be reduced by shielding as

much of the wiring as possible. This takes the form of a grounded
fine woven metal braid and/or metal foil surrounding a set of wires
and is built into cables used to connect experimental equipment.
Static electromagnetic waves are unable to penetrate a volume of
space entirely surrounded by a conducting shield and oscillating
electromagnetic waves undergo an exponential decay through the
shield [7].
All cable shielding and grounds of equipment should be con-
nected together to a single grounding point. Alternatively the cable
shielding is sometimes broken in a certain place to isolate two true
ground points from each other. If great care is not taken over this
then there may be more than one grounding point in the circuit. An
(electric) potential difference can arise between these points which
can cause large noise currents to flow and adversely affect the qual-
ity of the measured signal significantly.
Inductive cross-talk is noise resulting from the mutual induc-
tance of two or more closed circuits with paths located near each
other i.e. the changing magnetic field created by an alternating cur-
rent in one circuit creates spurious signals in a neighboring circuit
according to Faraday’s law V=d/dt. Figure 3.10 shows an arrange-
ment of wires that are subject to cross-talk. In the left wire a current
I with frequency wr flows. The current creates a magnetic field at
distance r from the wire at time t given by

Figure 3.10: Inductive cross-talk between wires in two closed circuits carrying AC signals.
 Making Measurements • 63

m0 I sin wr t
B r, t   (3.13)
2p r

The other two wires are part of a neighboring circuit and situated
parallel to the first wire a distance of r1 and r2 away. The separation
of these two wires (r2−r1) continues for length l. Ampere’s law can
be used to show the magnetic flux induced through this area of the
circuit is
  B r, t   dA
S
(3.14)
r2 m0 I sin wr t
l
  
r1 2p r
0
dzdr (3.15)
l m0 I sin wr t r2
 ln (3.16)
2p r r1

Using Faraday’s law shows the (electric) potential difference this

causes is proportional to wr and ln r2/r1
d
V  – (3.17)
dt
l m0 Iwr cos wr t r (3.18)
– ln 2
2p r1
To minimize EMFs induced in this way the frequency of the AC
current is kept low and the loop area between wires is kept to a mini-
mum by twisting both wires of a circuit together creating twisted
pairs. Another important reason to keep loop areas to a minimum
arises when measurements in a magnetic field are considered. Any
current carrying loop of wire in a magnetic field will be subject to a
torque (T=IAB) proportional to the current I, the loop area A and
the magnetic field B. As the current is AC, this will cause the loop to
vibrate and induce an EMF at the same frequency, wr as the current.
The best way to minimize this is to securely stick/tie down all wires
especially those which are very thin and light.
Lock-in amplifiers use a differential amplifier whose ideal out-
put is proportional to the (electric) potential difference between
its inputs i.e. the (electric) potential difference across the sample
V=Gain (V+ – V–). A real amplifier also has an output component
64 • Physics Lab Experiments

proportional to the actual (electric) potential differences


V  Gain   V – V–   10 – 6
 V  V–   . This additional compo-
 2 
nent is the common-mode (electric) potential difference. When
the (electric) potential difference measured by the lock-in is very
small this common-mode leak (electric) potential difference can
adversely affect the results. The best way to minimize the common-
mode (electric) potential difference is to balance the resistance of
each wire in the circuit with its pair.

3.15 SIGNAL GENERATOR

A signal generator produces a time varying EMF output. The

frequency is selected by choosing a range using the range dial and
then setting the frequency dial. The amplitude of the wave can be
increased or decreased using the amplitude/volume control. The
format of the wave can be chosen from a sinusoid, a square wave, a
triangular/saw tooth wave or the TTL output. Figure 3.11 shows an
example of a signal generator.

3.15.1 TTL

The TTL (Transistor-Transistor Logic) output provides a stan-

dard signal between 0 V and 5 V. A TTL output signal is defined as
“low” when between 0 V and 0.4 V with respect to the ground ter-
minal, and “high” when between 2.4 V and 5 V. A TTL input signal
is defined as “low” when between 0 V and 0.8 V with respect to the
ground terminal, and “high” when between 2 V and 5 V. These signal
are used to pass timing or triggering signals between different pieces
of equipment such as signal generators and oscilloscopes or current
sources and lock-in amplifiers.
 Making Measurements • 65

Figure 3.11: A signal generator.

3.16 OSCILLOSCOPES

Oscilloscopes are enormously useful pieces of equipment in

the laboratory provided their operation is understood. Figure 3.12
shows the front panel of an analogue or CRT oscilloscope and Fig-
ure 3.13 shows the front panel of a digital oscilloscope. The screen
on an oscilloscope can be loosely thought of as a graph. It is usually
marked with a graticule scale of large squares with length 1cm, as
well as smaller sub divisions. The horizontal axis is usually the time
and the vertical axis is the electric potential difference.
There are two inputs for BNC cables: Channel 1 (CH1) and
Channel 2 (CH2). BNC cables provide a firm connection which
does not break when pulled. The metal cap on the wire must be
pushed towards the instrument and twisted counterclockwise (clock-
wise) to disconnect (connect) the cable. Cables are usually coaxial
which means they have a core down which the signal travels. This is
66 • Physics Lab Experiments

Figure 3.12: An CRT oscilloscope.

surrounded by an insulating wrapper and then anther conducting

wrapper which may be foil or braid and acts as an electrical shield.
Finally this is surrounded by another insulating sheet.
The following list gives a brief description about the function
and use of common oscilloscope buttons.
● “CH1 Volts/Div” and “CH2 Volts/Div” also sometimes
called the “Vertical Gain”. There is one dial per channel,
which controls the vertical amplitude of the waveform.
The settings are usually given in terms of Volts per cm. So
decreasing the Volts per cm, increases the gain and the
amplitude of the waveform on the screen. The vertical gain
is normally set so that the waveform fills the screen from top
to bottom without going off the screen. There are three com-
mon sources of confusion in checking amplitude measure-
ments are as expected.
● AC (electric) potential differences are usually discussed in
terms of RMS (root mean square) values. Oscilloscopes dis-
play peak values which are 2 = 1.41 times larger than RMS
values.
● Sometimes there are calibration controls in the center of the
Volts/Div and timebase dials. These should be in a normal
position indicated by a “locking” felt at the dial is rotated.
 Making Measurements • 67

● Sometimes special probes are used on an oscilloscope

which scale the signal by a factor of 10 or 100. This will
be printed on the probes attached to the CH1 and CH2
inputs.
● Some oscilloscopes have a button to magnify the x and/or y
axis by a factor of 10 independently of the volts/div or time-
base controls.

● Vertical Position This controls the vertical position of the

waveform on the screen. When no signal is present the trace
on the screen is usually adjusted so that it sits on a convenient
graticule marking on the screen so that amplitudes above and
below zero can be easily measured. When taking amplitude
measurements it is sometimes convenient to adjust the verti-
cal position to line up the peak and trough of the waveform
with the graticule lines.
● Timebase The timebase or SEC/DIV dial controls the
speed at which the screen is scanned. The settings are usu-
ally given in terms of the number of seconds per cm. The
frequency of the waveform can be calculated from this by
taking one divided by the time period for one wavelength.
For example if the time base setting is 0.01 seconds/cm
and one complete wavelength takes 5 cm, the period is
0.05 seconds and the frequency is 1/0.05 = 20 Hz. Similar
to the vertical gain, the timebase is set so that the wave-
form fills the screen with a small number of periods from
right to left.
● Horizontal Position This is similar to the vertical position
control, but changes the horizontal position. This can be
adjusted when taking time measurements so that the wave-
form lines up well with the graticule lines.
● Autoset Some modern oscilloscopes have an autoset button.
When this is pressed, the CH1 gain, CH2 gain and the time-
base are automatically set to appropriate values. This is an
especially useful feature to have as it saves a great to deal of
wasted time when oscilloscope settings are unfamiliar or the
frequency/magnitude of the signal is unknown.
68 • Physics Lab Experiments

● Trigger The trigger setting controls the point at which

the waveform begins to appear on the left hand side of the
screen. Each time the waveform reaches the right hand side
of the screen the oscilloscope will pause and wait until a cer-
tain condition is met before starting again at the left hand
side of the screen. For periodic functions this provides a syn-
chronised waveform which does not drift across the screen
and it enables randomly occurring pulses to be displayed
easily. The trigger level control sets the (electric) potential
difference which the input has to reach before the display
of the waveform is restarted. Other trigger related buttons
can control when the trigger is based on channel 1 or chan-
nel 2 or whether it is based upon the rising or falling edge of
the waveform. The trigger can also be set to external. This
triggers the oscilloscope based upon a pulse from an exter-
nal source such as the TTL output from a signal generator or
lock-in amplifier.
● Mode There are usually a series of buttons or a dial which
sets what is displayed on the screen (in Figure 3.12 they are
down the right hand side of the screen). This allows a choice
between displaying only channel 1 (CH1), only channel 2
(CH2), both channel 1 and 2 at the same time (DUAL), the
sum on channel 1 and channel 2 (ADD). The XY mode dis-
plays the CH1 input as the x coordinate and the CH2 input as
the Y coordinate. If two different AC signals (at frequencies f1
and f2) are connected to CH1 and CH2 when the oscilloscope
is in XY mode, interesting patterns called Lissajous figures
appear when f2 = nf1.
● Coupling The GND/DC/AC buttons for CH1 and CH2 al-
ter the coupling between the signal input and the input to
the amplifier. The DC setting means that the input is con-
nected directly to the amplifier. The AC setting means that
a capacitor is connected between the input and the amplifier
so that DC (electric) potential differences are blocked. The
GND setting means that the amplifier input is connected
to the ground (0 V) which allows a check of the position of
0 V on the display. The DC position should be appropriate in
most cases.
 Making Measurements • 69

● Measurements More modern, digital oscilloscopes have

horizontal and vertical cursors or bars which can be displayed
on the screen. Usually the button is labeled “Measure”. The
cursors can be moved and the screen displays the (electric)
potential difference between the horizonal cursors and the
time period between the vertical cursors. It is also possible to
get the oscilloscope to calculate and display the period, fre-
quency and amplitude of the signals.

To begin using an oscilloscope first find and operate the power but-
ton on the front panel. It is then usual for a power LED to turn on
and perhaps a number of other LEDs. If nothing happens, check for
an additional power switch on the back of the oscilloscope and that
the mains lead is properly connected and that the mains is turned
on.
Wait for the beam to be displayed. An analogue oscilloscope may
take a moment or two to turn on and a digital one will take longer as
the software starts up. If the trace is not displayed on the screen then
check the trigger level dial is set to the center and the horizontal and
vertical position controls are set to the center.
Next connect the signal via the BNC connector on the front
panel to at least one channel. The coupling should be set to DC in
most cases and the mode set to either CH1, CH2, or DUAL depend-
ing on where the input signal(s) are connected. Set the vertical gain
and timebase dials to an appropriate setting so that the waveform
occupies a large part

Figure 3.13: A digital oscilloscope.

70 • Physics Lab Experiments

of the screen. If no signal appears, try pressing the autoset button

if the oscilloscope has one otherwise the BNC cables and any other
cables should be checked. If there appears to be poor connection,
then the cables should be swapped for a different set. If this fails
the oscilloscope could be connected to a signal generator which is
known to be working.
Finally the trigger should be set to one of the channels with a
signal and the level control can be adjusted along with the rising or
falling edge selection so that the best waveform is displayed.

3.17 LIGHT GATES

Light gates are very useful for measuring time, speed/velocity

and acceleration of moving objects. To simplify and improve the flow
of the language, in this section it will be assumed the object is a toy
car. Of course, in reality it may be any other suitable moving object.
It is vital to point out that light gates are timers. The only measure-
ment they make is the time for which a light beam (usually infrared)
is broken. Further processing and thoughtful setup is required to
find the velocity and acceleration using either one or two light gates.
Often, the light gate control box or computer will do much of the
processing, but will initially need to be programmed with a number
of distance measurements.

3.17.1 VELOCITY WITH ONE LIGHT GATE

Usually a small, vertical piece of card of known length would be

fixed to the top of the car. The velocity of the car at a given point is
found by placing the light gate at the point and dividing the length
of the card by the time taken for the card to travel through the light
gate. If a piece of card is not used, it must be clear the exact length
of the car which is breaking the light beam.
 Making Measurements • 71

3.17.2 VELOCITY WITH TWO LIGHT GATES

In this case a piece of card is not necessary and the method

may be more suitable where card can not easily be used. The two
light gates are setup a known short distance apart. The light gates
are setup to measure the time between the gates, so that the timer
is started as the car travels through gate A and is stopped when the
car travels through gate B. The velocity is found by dividing the
distance between the light gates by the measured time. This gives
an average velocity over the distance between the two light gates
and is similar to how average speed checks on roadways in the US
work.

3.17.3 ACCELERATION WITH TWO LIGHT GATES

Two velocity measurements taken a known time apart are

needed. Each of these can be taken in a similar way to the measure-
ment of the velocity with one light gate. The average acceleration of
the car between the two light gates is found by diving the difference
in the two velocity measurements by the duration of time between
them.

3.17.4 ACCELERATION WITH ONE LIGHT GATE

To measure the acceleration with just one light gate, two veloc-
ity measurements are still needed. These can be taken using a “U”
shaped card. The first velocity is measured using the first card to
break the beam and the second velocity is measured using the sec-
ond card to break the beam. Here it is important to measure the
width of each of the vertical sections of the “U”. The acceleration is
calculated in a similar way by diving the difference in the two veloci-
ties by the duration of time between them.
72 • Physics Lab Experiments

3.18 PASCO SCIENCE WORKSHOP INTERFACE

PASCO Scientific produce a range of interface boxes to connect

a measurement devices to a computer for data logging. One example
of this is the Science Workshop 500 Interface see: Figure 3.14. This
box connects to the mains via a transformer and to the computer via
a USB to serial converter. A large selection of sensors are available
such as: photogate, ultra sound motion sensor, (electric) potential
difference, temperature, pressure, and force. These are connected
to digital or analogue channels inputs on the interface.
Specially designed “PASCO DataStudio” software is used to
record the data into the computer and is available for Mac and Win-
dows, although it has now been superseded by “PASCO Capstone”.
Plug in the USB wire between the interface box and the com-
puter. Select “File, New Activity”. Then pick “Create Experiment”.
The interface should be detected and an “Experiment Setup” win-
dow should appear. Plug the sensor into the interface box. Select the
correct sensor from the list by double clicking on it: see Figure 3.15.

Figure 3.14: A PASCO Science WorkShop 500 Interface box connected to a photogate.
 Making Measurements • 73

For example, choose the photogate from the “Science Workshop

Digital Sensors” drop down. Select the measurement required, for
example “Velocity in Gate, Ch1”.
Choose the method required for the display by clicking and
dragging a display type from the “Displays” panel at the bottom left
to the measurement in the “Data” panel at the top left. For example,
choose “Digits”.
To record data, press the “Start” button in the menu bar at
the top of the screen. The “Digits” display will update each time
the light beam is interrupted. Data taking can be stopped using the
“Stop” button. See Figure 3.16.
Alternatively, setup an ultrasound motion sensor to measure
position and select a “Graph” from the “Displays” panel. Use the
“Motion Sensor” tab in the “Experiment Setup” window to calibrate
the sensor. Position the sensor a “Standard Distance” from a large
solid surface such as a wall, floor, or a large book. Then press the
“Set Sensor Distance = Standard Distance” button. Record the data
by using the “Start” button as before. See Figure 3.17.

Figure 3.15: Choosing a sensor in PASCO DataStudio.

74 • Physics Lab Experiments

A table of values can be saved by dragging the “Table” option

from the “Displays” panel up to the “Data” panel even after the data
has been recorded. The data can be saved for use later or in other
software by choosing “Export Data” from the “File” menu.

3.19 DATA HARVEST

Data Harvest QAdvanced dataloggers give reliable datalogging

which is quick and easy to setup, see Figure 3.18. They can be used
in either a standalone mode or when connected to a computer. When
connected by a USB cable to a (Windows) computer the “Data Har-
vest EasySense” software can be used to record, graph, and save
data. With a sensor connected to one of the inputs on the datalog-
ger, when the program opens choose “EasyLog” and then press the
“Start/Stop” button. Live data will be displayed on the graph. See

Figure 3.16: Using PASCO DataStudio to measure the velocity in a photogate.

 Making Measurements • 75

Figure 3.17: Using PASCO DataStudio to produce a distance time graph.

Figure 3.18: A Data Harvest QAdvanced Datalogger with a

Light Level probe connected to input 1.
76 • Physics Lab Experiments

Figure 3.19: A screen shot of the Data Harvest EasySense software

showing a graph of Light Level against time.

Figure 3.19. Alternatively, in a stand alone mode, the results are dis-
played on the built in screen or recorded and saved to the datalogger.
For example to setup a velocity measurement using a single light
gate, connect the light gate to the input labeled A. Press any button
to turn the device on. The arrow buttons can be used to navigate
through different menu options. Navigate to “Time & Motion” and
press the green arrow button. Navigate to “Speed” and press the
green arrow button. Navigate to “Speed at A” and press the green
arrow button. Choose the “1 Interrupt card” option and press the
green arrow button. Finally navigate to the length of the interrupt
card being used and press the green button. The datalogger will dis-
play the latest speed as well as an average speed.
To use a Light Level probe, connect this to input 1. Navigate to
“Meter” and press the green arrow button. The display will show a
reading updated a few times per second.
Use the red square button to go back to the previous menu. Nav-
igate to “Logging” and press the green arrow. Use the arrow keys
 Making Measurements • 77

set a “Duration” and “Interval” for the measurements and choose a

start time. Logged data is saved to the datalogger, this can be down-
loaded to the “Data Harvest EasySense” software on the computer
for graphing or analysis. When the program starts, choose “Remote”
and then “Retrieve Remote”. Select the data to collect from the dat-
alogger and click the “Retrieve” button. The data is now displayed as
a graph on the computer: see Figure 3.19.

3.20 PRESSURE GAUGES

There are a number of pressure gauges which may be met. Each

is based upon a different mechanical or electrical way of detecting
pressure and is sometimes only suitable for certain pressure ranges
or a certain resolution. A table of different pressure units was given
in Section 2.5.1.
A basic pressure gauge can be made using a “U” shaped tube which
is half full of a liquid (such as water or mercury). One side of the U tube
is connected to a known reference pressure P1 such as atmospheric
pressure or a vacuum and the other is connected to the pressure region
to be measured P2. The difference in liquid level, h, between the two
sides of the “U” tube is related to the pressure difference:

P1 – 2  r gh (3.19)

where g is the acceleration due to gravity and r is the density of the

liquid.
Mechanical gauges measure pressure by monitoring an element
which flexes or bends due to a pressure difference across it. The ele-
ment which may be a Bourdon tube, a diaphragm or another similar
device which causes a pointer to move against a calibrated scale.
A Bourdon tube is a flattened tube which tries to regain its circu-
lar cross section when it is pressurized. This effect can be ampli-
fied sufficiently to turn a pointer by forming the tube into a “C”
shape or a coil. A diaphragm is a flexible membrane which sepa-
rates areas of different pressures. The change in position of the
78 • Physics Lab Experiments

membrane is reproducible, related to the pressure difference across

the membrane and can be used to move a pointer.
The mechanical gauges may be read electronically by linking the
Bourdon tube to a piezoelectric strain gauge or the diaphragm to a
capacitance sensor.
A Pirani gauge is a common electronic gauge useful for measur-
ing pressures between 0.1 Pa and 1000 Pa. It consists of a metal
wire which is heated by a current flowing through it. Thermal equi-
librium of the wire is achieved due to a cooling effect of the gas
surrounding the wire. If the gas pressure decreases there will be
fewer collisions of gas molecules with the wire meaning its tempera-
ture will increase. The increase in temperature of the wire causes its
resistance to increase. Measurements of the current flowing through
the wire and the electric potential difference across the wire are
used to find its resistance. A calibration table or mechanism is then
used to convert the resistance of the wire in to pressure.
Ionization pressure gauges are more sensitive and measure pres-
sures in the range 10−8 Pa to 10−1Pa. Free electrons are generated
which ionize gas molecules. These gas molecules are attracted to an
oppositely charged electrode. The small current between the posi-
tive and negative electrodes is proportional to the rate of ionization.
Lower density (therefore lower pressure) gases produce less ions
and so the current is linked to the pressure. However, the number
of ions produced can be dependent on the gas so a good calibration
for the appropriate gas is necessary. There are two subtypes: hot
cathode gauges and cold cathode gauges. Hot cathode gauges use
a heated filament to produce electrons via a process called therm-
ionic emission. The most common of these is called a Bayard-Alpert
gauge. A cold cathode gauge produces electrons by high EMF dis-
charge. The most common of these are called a Penning gauge or an
Inverted Magnetron.

3.21 TEMPERATURE MEASUREMENT

There are a board range of methods for measuring temperature

in the laboratory. The chosen method will be different depending on
 Making Measurements • 79

the temperature range to be measured, the accuracy and precision

required and the conditions under which the measurement must be
taken. Temperature can be measured in Kelvin (K), Celsius (°C) or
Fahrenheit (°F). Temperatures can be converted from one scale to
another using:

TF  1.8TC  32 (3.20)
TK  TC  273.15 (3.21)

3.21.1 NEAR ROOM TEMPERATURE

Mercury and Alcohol Thermometers

A glass column is attached to a small reservoir or bulb at one
end. This is filled with a liquid whose volume depends significantly
on temperature. The two most common liquids are mercury and
alcohol. Mercury thermometers are becoming rarer as government
rules try to reduce the amount of mercury in the environment and
as of 2012 their sale to (although not use by) the public is prohibited.
Mercury thermometers have been replaced by colored chemical
thermometers. As the liquids increase in temperature their volume
increases and they expand up the glass column.
Mercury thermometers are limited by the temperature range
in which mercury is a liquid (39°C up to 357°C). Similarly ethanol
thermometers only work when ethanol is a liquid between 114°C
and 78°C. To enable mercury free thermometers to be used above
78°C other chemicals can be used such as isoamyl acetate, toluene,
or kerosene which have higher boiling points.
IR Thermometers
An infrared thermometer infers the temperature of an object
from the thermal radiation it emits. By knowing the amount of infra-
red radiation emitted by the object and its emissivity, the object’s
temperature can be determined.
A black body is an idealized object which absorbs all incident
electromagnetic radiation no matter what the angle of incidence or
80 • Physics Lab Experiments

frequency. A black body emits electromagnetic radiation, of intensity,

I, where I(f,T), is the amount of energy per unit surface area per unit
time per unit solid angle emitted at a frequency, f, by a black body at
temperature T. This is Planck’s Law:

2h f 3 1
I( f , T)  (3.22)
c2 ehf / kT – 1

where c, is the speed of light, h is Planck’s constant and k, is the

Boltzmann constant. Integrating this over solid angle and all fre-
quencies gives the Stefan-Boltzmann Law:

P  AesT 4. (3.23)

where P is the total power radiated by the object, A is the surface

area of the object, e is the emissivity of the object (e = 1 for a per-
fect black body and e < 1 for all other objects),  is the Stephan-
Boltazman constant which is 25k4 /15c2h3 = 5.67 ×10− 8 Wm −2 K–4
and T is the temperature of the object. This means that the amount
of emitted radiation is determined by only the temperature of the
object.
Infrared thermometers are useful as they can make non-contact
measurements in environments where other thermometers would
not work or would be difficult to implement such as in a vacuum, at
very high temperatures, in electromagnetic fields or where access
is difficult. Sometimes they have a built in laser spot used to aim
the thermometer indicating approximately from where the emitted
radiation is being measured.
A infrared thermometer may give an inaccurate reading when it
is not correctly aimed, when the emissivity is only given an approxi-
mate value for a particular surface/object or if stray radiation is
picked up from nearby objects (usually at a higher temperature).
Thermocouples
Any conductor which is subjected to a thermal gradient (between
the temperature of a sample and known reference temperature) will
generate a small (electric) potential difference across its ends. This is
 Making Measurements • 81

known as the thermoelectric effect or Seebeck effect. It can be mea-

sured using a voltmeter.
In order to form a complete circuit to measure this (electric)
potential difference a further conductor must be connected to the
sample end of the first conductor. This further conductor will be
subjected to the same temperature gradient which will develop its
own (electric) potential difference which will oppose and exactly
cancel the original (electric) potential difference. However, the
magnitude of the effect depends on the metal from which each con-
ductor is made. Thus, using two dissimilar metals results in a small,
but measurable, net (electric) potential difference. The measured
(electric) potential difference increases with temperature, and is
typically between 1 and 70 µV/°C.
Since the (electric) potential difference is generated along the
portion of the length of the two dissimilar metals that is subjected to
a temperature gradient and both lengths of dissimilar metals experi-
ence the same temperature gradient, the end result is a measure-
ment of the difference in temperature between the thermocouple
junction and the known reference temperature.
Thermocouples can be purchased according to standard speci-
fications denoted by letters. Different types are best suited for dif-
ferent applications. They are usually selected on the basis of the
temperature range, sensitivity needed, how inert the metals are in
the measurement environment and sensitivity to magnetic fields.

Table 3.4: Thermocouples, Compositions, Temperature Ranges, and Sensitivities.

82 • Physics Lab Experiments

Table 3.4 gives the properties of some common thermocouples: K is

the most commonly used, and E is not affected magnetic.

3.21.2 LOW TEMPERATURES

A range of methods have been developed to allow measurement

of low temperatures. Each method has associated advantages and
disadvantages. It is likely that one or more of these methods will be
met but the one in use will depend upon the experimental require-
ments and environment.
Gas Thermometry
With this method it is difficult to obtain a high accuracy due
to the corrections needed for inaccuracies such as the volume of
connecting tubing, contraction of the bulb, and pressure corrections
due to non-ideality of the gas. The pressure measurements become
more accurate when diaphragms are used to separate the gas in the
bulb and gas in the pressure gauge. Electronics can then be used to
detect the movement in the diaphragm. This method also has the
advantages that it is unaffected by magnetic fields and that it is easy
to measure differential temperatures. Typically 3-He or 4-He gas
would be used since they have a very low boiling point.
Resistance Thermometry
Resistance thermometry depends upon the variation of resis-
tance with temperature. This method is much easier than the
gas thermometry to perform and is reliable down to a lower tem-
perature. A good material would have a (relatively) linear depen-
dence of resistance on temperature, be easily obtainable in high
purity, chemically inert and with a stable resistance. Metals such
as platinum (Pt) are suitable. The possible resolution when using
Pt decreases below about 10 K so materials such as Rhodium Iron
(RhFe) are more useful. RhFe also has the advantage that is has
a large and positive temperature coefficient below 30 K. Another
useful material is Arsenic (As) doped Germanium (Ge). Thin Film
Resistors using certain materials are prone to the effect of magnetic
fields whereas carbon-glass thermometers are only slightly affected
 Making Measurements • 83

by a magnetic field. Thick film resistors using Ruthenium Oxide

(RuO2) have also been used due to their low cost, ease of use, ease
of lowering their temperature due to a low heat capacity and the
possibility of predicting their low temperature resistance from that
at room temperature.
Electronic Thermometry
Electronic thermometry has a wide temperature range, a
higher sensitivity compared to other methods, simpler operation
and a more linear V verses T characteristic than most resistance
thermometers. Silicone (Si) junctions make the most stable and
reproducible thermometers, but they have high magnetic field
dependence. A Gadolinium-Aluminium-Arsenic (GaAlAs) semi-
conductor has a much reduced dependence on magnetic field. They
are inexpensive, but electrical noise limits currents to above 10 µA
so heat dissipation becomes a problem at very low temperatures.
Carbon and Germanium based semiconductors show a significant
hysteresis effect when cooled again after being allowed to reach
room temperature.
Thermocouples
Low temperatures can also successfully be measured using
thermocouples. Stable, temperature sensitive alloys with a strongly
temperature dependent magnetic moment must be used eg. those
of Cu or Au. Great care is needed to get an accuracy of less than
1 % especially at low temperatures since the thermocouples become
much less sensitive as the temperature drops.
Capacitance Thermometry
Capacitance thermometers, primarily using SiO2-OH, have
been successfully developed and have become popular due to their
insensitivity to magnetic fields. The readings are stable and have a
constant sensitivity right down to just above 0 K. However, after they
are cycled to room temperature they show a significant hysteresis
and drift effects.
Calibration
Thermometers operating at low temperatures can be calibrated
in a number of ways. The first is by extrapolation of an existing
scale down to lower temperatures. The second, is by direct mea-
84 • Physics Lab Experiments

surement of a few known points such as in liquid helium (4.2 K),

liquid nitrogen (77 K) and ice water (273 K). Mathematics can be
used to interpolate temperatures between these. Alternatively a
comparison with a primary (previously calibrated) thermometer
can be made.
These methods can fail due to unknown additional temperature
dependence at lower temperatures or due to unexpected and there-
fore uncorrectable systematic errors.

3.21.3 HIGH TEMPERATURES

High temperatures can be measured without contact using an

infrared thermometer. Resistance thermometry may also be appro-
priate for contact measurements in a similar way as they are used
at low temperatures. Platinum is particularly appropriate due to a
linear and well defined resistance, its resistance to oxidation or other
degradation in the high temperature environment and its relatively
high melting point. A further method in common use is the thermo-
couple.

3.22 PID AND TEMPERATURE CONTROL

A PID controller is able to change and hold a parameter at a par-

ticular value called the setpoint. It consists of a control loop mecha-
nism with feedback and is usually realized in an electronic circuit. It is
commonly found in temperature controllers where a temperature can
be set, monitored using a thermometer and increased using heater.
Cooling usually occurs due to heat transfer to the surrounding.
A PID controller is named after its three correcting terms: the
proportional, integral and derivative. These are summed to calculate
the output of the PID controller. Defining x(t) as the controller out-
put (which in the case of a temperature controller may be applied to
the heater), the PID algorithm is:
 Making Measurements • 85

t d
x  t   Pe t   I  e t dt  D dt e t 
0
(3.24)

where P is the proportional gain constant, I is the integral gain con-

stant, D is the derivative gain constant, e is the error (the difference
between the set point and the present value), t is the time and  is
the variable of integration.
The proportional term produces an output value that is pro-
portional to the current error. The proportional output is found by
multiplying the error by the proportional gain constant, P. A high
proportional gain results in a large change in the output, x(t) for a
given change in the error. A small proportional gain results in a small
change to the output for the same change in error making the con-
troller less responsive or less sensitive.
The integral term produces an output value that is propor-
tional to both the magnitude of the error and the duration of the
error. It is the sum of the instantaneous error over time and gives
the accumulated offset that should have been corrected previ-
ously. The accumulated error is multiplied by the integral gain
constant I to give the integral output. The integral term acceler-
ates the movement of the process towards the setpoint. How-
ever, since the integral term responds to accumulated errors from
the past, it can cause the present value to overshoot the setpoint
value.
The derivative term produces an output value that is propor-
tional to the slope of the error over time. Is is found by multiplying
the rate of change of error with time by the derivative gain constant,
D. The derivative term slows the rate of change of the controller
output. Derivative control is used to reduce the magnitude of the
overshoot produced by the integral component. However, the deriv-
ative term slows the response of the controller. Also, differentiation
of a signal amplifies noise and thus this term in the controller is
highly sensitive to noise in the error term.
In order for a system to reach its setpoint quickly the PID values
must be adjusted until their optimum value is found. This can be
done manually: the I and D values are set to zero. The P value is
increased until the present value oscillates. P should then be set to
86 • Physics Lab Experiments

Table 3.5: Effect of changing PID values.

half the value which caused the oscillation. The present value will
then overshoot the setpoint. The I value can then be increased from
zero until the offset is corrected in a reasonable time. Finally the D
value can be increased so that any instability in the system is cor-
rected in a reasonable time.
Table 3.5 gives the effect of changing one of the PID constants
independently: An alternative to setting the PID gain constants
manually is to follow one of a number of pre-defined algorithms for
setting them such as the Ziegler–Nichols method [8]. As before the I
and D gain constants are set to zero. The P gain constant is increased
until it reaches the ultimate gain PU at which the output of the loop
oscillates with a constant amplitude. The values of PU and TU (where
TU is the time period of oscillation) are used to set the values of the
PID gain constants using the relationships shown in Table 3.6 [9].

3.23 KAYE AND LABY TABLES

A vital part of any Physics Laboratory has been a book

of physical constants and material properties. It has allowed

Table 3.6: Relationships for setting PID values.

 Making Measurements • 87

physicists to look up quantities they need for data analysis or to

check the accuracy of values they have measured. The standard
book is Kaye and Laby Tables of Physical and Chemical Constants.
This can be purchased as a printed book, but is also now freely
available online at http://www.kayelaby.npl.co.uk/. This book and
website give a wealth of information on a wide range of topics.
Naming only three as examples: nuclear decay chains for every
radio active isotope, the pressure dependence of the boiling point
of organic compounds and the velocity of sound in a range of gases.

3.24 MICROSCOPES

3.24.1 Optical Microscopes

Optical microscopes can be a valuable tool for enlarging small
samples. The wavelength of visible light is in the range 400 nm–700
nm, so distances smaller than this can not be resolved by an optical
microscope, no matter how good the quality.
To view samples under an optical microscope a good light source
is needed to illuminate the sample. Typically optical microscopes fall
into two categories. Most common are transmission microscopes,
which illuminate a sample primarily from beneath and rely on light
transmitted through the sample. More useful in physics is a reflec-
tion microscope which usually have larger lenses to allow the collec-
tion of more light. They rely on the sample being brightly lit from
above and collect light reflected from the sample.
Most microscopes will have position adjustable eyepieces which
should be set to suit the user. Both the eyepiece and objective lens
will have distinct magnifications, the total magnification is given
by their product. Better quality microscopes will have a range of
different magnifications to choose from selecting the appropriate
objective lens.
As well as the limit imposed by the wavelength of light, optical
microscopes suffer from various aberations which reduce the quality
of the images. Chromatic aberations occur when light of different
colors (different wavelengths) is not focused at the same point. This
is due to an effect called dispersion of the lens which arises because
88 • Physics Lab Experiments

Figure 3.20: Illustration of chromatic aberation of a lens. Red light (the dashed line) is
refracted less than blue light (the dotted line).

the lens has a decreasing refractive index for increasing wavelengths

of light. Figure 3.20 illustrates the effect of a chromatic aberation.
Chromatic aberation produces fringes of color along boundaries
between light and dark parts of the image. A simple method of reduc-
ing chromatic aberation is to make the focal length as long as possible
or use monochromatic (a single wavelength) of light. More modern
methods involve the use of low dispersion glass or the production of
achromatic lenses using layers of materials with different dispersions.
Spherical aberations result from the use of spherical lenses.
These are cheaper and easier to produce than aspherical lenses.
Light which strikes the lens further from the center is refracted too
much compared to light which strikes the lens nearer the center.
Figure 3.21 illustrates the effect of spherical aberations. Spherical
aberation can be reduced by making aspherical lenses or by combin-
ing a number of spherical lenses.

Figure 3.21: Illustration of spherical aberation of a lens.

 Making Measurements • 89

Astigmatism occurs when light rays which propagate in two perpen-

dicular planes have different foci. They typically arise as an artifact
of the manufacturing process.

3.24.2 TRANSMISSION ELECTRON MICROSCOPE

To observe distances smaller than the limit of an optical micro-

scope electrons can be used as they have a shorter wavelength:
typically 3.4 pm for 120 kV electrons. Samples must be electrically
conductive. Insulating samples are first covered with a thin film of
metal such as gold.
Electrons are usually created by thermionic emission and then
accelerated by passing through a high potential difference. The
higher the potential difference, the higher the energy and the
smaller the wavelength of the electrons, meaning smaller features
can be resolved.
The lenses in a transmission electron microscope (TEM) are
made using electromagnets. Glass lenses would not work as elec-
trons do not penetrate far into matter (hence very thin samples are
needed to observe the diffracted beam). Permanent magnets would
be equally useless because the image could not be focused. Electro-
magnets allow the lens currents to be varied and so allow the image
to be focused onto a screen. A vacuum is needed inside the column
of the microscope because electrons would collide with air mole-
cules rapidly destroying the beam.
When the electron beam hits the sample some electrons are scat-
tered and some are transmitted. If the transmitted beam is selected
using the correct size aperture a bright field image is formed. An
aperture is usually a platinum metal disk with a hole in the center
placed in the path of the electrons. Since the scattered electrons are
stopped by the objective aperture the amount of scattering deter-
mines the contrast in the bright field image. Where many electrons
are scattered the image will look dark, where few are scattered the
image will be lighter. A dark field image is formed when the scat-
tered beam is selected using an aperture designed to block the
direct beam. An advantage of dark field images is the much higher
90 • Physics Lab Experiments

Figure 3.22: Left: Electron diffraction pattern from Gold. Right: A dark field image of
islands of Gold.

contrast which can be obtained, although a large electron flux is

needed which may damage the sample.

3.24.3 OTHER MICROSCOPES

Alternatively a scanning electron microscope (SEM), scan-

ning tunnelling microscope (STM) or atomic force microscope
(AFM) may be used. It is unlikely these microscopes will be
encountered in the undergraduate laboratory, however it is still
useful to have a brief overview of their operation. It should be
noted that images taken with any of these microscopes are often
given false colors. Since many of the features imaged are less
than 400 nm (the wavelength of violet light) they can’t actually
have a color in the traditional sense as they are not capable of
reflecting visible light.
A scanning electron microscope (SEM) works in a similar way to
a TEM except a narrow focused beam of electrons is scanned across
the surface of the sample producing a image with a large depth of
field giving a three dimensional appearance and a resolution down
to about 1 nm. As with a transmission electron microscope, tradi-
tional SEM require samples to be placed in a high vacuum and be
electrically conductive.
 Making Measurements • 91

There are a range of methods of making measurements: trans-

mitted electrons can be monitored or back scattered (reflected) elec-
trons can be collected which allows areas with different chemical
compositions to be identified as elements with high atomic numbers
back scatter electrons more than elements with low atomic num-
bers. Alternatively characteristic X-rays can be detected, they are
emitted when the electron beam removes an electron from the inner
shell of an atom. This causes a higher energy electron to drop down
to take the place of the ejected electron, releasing energy in the
form of an X-ray. Characteristic X-rays are used to identify elements
present and their relative proportions in the sample. Acceleration
and scintillation based detection of secondary electrons allowed the
development of an environmental scanning electron microscope
which allows insulating samples to be imaged in low pressure gas
or water vapor. Secondary electrons are those removed from the 1s
(or k) orbital in atoms in the sample by inelastic scattering with elec-
trons from the electron beam.
A scanning tunneling microscope (STM) can give a resolution of
around 0.1 nm and a depth resolution of 0.01 nm. This is sufficient
to image and even move individual atoms. Little sample preparation
is needed: the STM works in vacuum, air or liquid over a wide range
of temperatures. A very sharp conducting tip is positioned above the
sample and scanned across the surface to build up the image pixel
by pixel. A potential difference (or bias voltage) is applied between
the sample and the tip which allows electrons to quantum mechani-
cally tunnel between them. The resulting tunneling current can be
measured and used to produce an image of the hight/relief of the
sample. The challenge with STM is to have a very clean sample sur-
face, the tip needs to be very sharp with just a single atom at the tip
and there needs to be very good vibration suppression in the equip-
ment and the building.
An atomic force microscope (AFM), similar to the STM, also has
a sharp tip which is scanned across the surface of a sample. How-
ever with an AFM the tip is on the end of a cantilever. As the tip is
brought close to the surface of the sample forces between the tip
and the surface cause the cantilever to deflect a small amount. This
is typically measured using a laser which is reflected from the top
surface of the cantilever. This “optical lever’ allows a much bigger
92 • Physics Lab Experiments

deflection to occur in the laser which can more easily be measured.

Usually there is a feedback mechanism which adjusts the height of
the sample to avoid the tip colliding with the surface. Images pro-
duced with an AFM give a three dimensional view of the surface
of the sample. The sample needs no special treatment and can be
imaged at room temperature in either air or liquid.

3.25 SPECTROSCOPES

A spectroscope or spectrometer takes light from an object and

splits it into its component colors/energies. Typically light will be
brought to a focus on a slit at the entrance to the spectroscope. Light
which passes through the slit falls onto a mirror which collimates the
light i.e. makes the light rays parallel. This light is then directed to
a diffraction grating. The angle at which the intensity minima and
maxima of the interference/diffraction pattern occur depends on the
wavelength of the light. The result is that the incident light is spread
out. If white light entered the spectroscope the spectrum would be
a rainbow pattern. If a sodium lamp is used, this only emits two spe-
cific wavelengths of orange light so there would be an emission spec-
trum showing two narrow orange lines against a dark background. A
camera may be used to image the resulting spectrum and measure-
ments may be taken to identify the wavelength of the emission or
absorption lines.

3.26 COMPUTERS

3.26.1 Plugs, Sockets and Connectors

Table 3.7 gives images of a wide range of connectors which
might be frequently used in the laboratory. It is does not constitute a
complete list: even for a type of connector which is listed there may
be other sizes or shapes of connector not shown. These images may
be helpful in identifying unknown connectors.
 Making Measurements • 93

Name Male (Plugs) Female (Sockets)

Mains

Mains

Mains

3 Phase
Mains

VGA

DVI

S-Video
94 • Physics Lab Experiments

Composite

HDMI

3.5mm
Audio

USB Type A

USB Type B

mini USB
 Making Measurements • 95

USB 3

miniUSB 3

FireWire
(IEEE1394)

mini
FireWire

BNC

9pin Serial
(RS-232)
96 • Physics Lab Experiments

25pin Serial
(RS-232)

Parallel

Parallel

GPIB
(IEEE488)

PS-2

Ethernet
(RJ45)

Modem
(RJ11)

Table 3.7: Common plugs and sockets.

 Making Measurements • 97

3.26.2 COMPUTER DATA LOGGING

This section provides a brief tutorial on setting up Delphi,

LabView and MATLAB to record experimental data. It is likely that
any computer used in the undergraduate laboratory will already be
setup. However this section aims to give an overview of what is going
on “behind the scenes” and provide a starting point for setting a
computer at a later point in time.
Instruments can communicate with a computer via a number
of different connection interfaces. This simply involves connecting
a cable from the instrument to a computer. Old interfaces include
parallel and serial ports, but a more modern method is by USB. Fre-
quently on modern computers serial ports are becoming obsolete
— although they can often be purchased on separate PCI cards or
as a USB adapter.
More specialist scientific equipment often connects via ethernet
(standard network cables) or GPIB (which requires a special adapter
card inserted inside a computer or a USB converter).
To use a GPIB interface a GPIB PCI card (perhaps manufac-
tured by Measurement Computing) needs to be installed into a stan-
dard Windows PC and the drivers correctly setup. A GPIB cable
is run from the back of the card on the PC to the first instrument.
Up to 14 further instruments can be connected in a daisy chain
with a single cable linking an instrument to the previous one. Each
instrument requires a unique address which is set in the configu-
ration options on the instrument itself. (Addresses can be chosen
regardless of where the instrument is located in the chain and also
of which other address are chosen provided each is unique). The
card driver installation on a Windows computer creates a shortcut
called CBCONF32 in the Start Menu. This program is used to link
the GPIB address of the instrument to a short text handle which is
used to refer to the instrument in the program code. Any other set-
tings required can be configured using this program too, although
the defaults are sufficient for most instruments.
Delphi
The Delphi programing environment needs to be installed on
the PC. Each instrument requires specific text commands sent over
98 • Physics Lab Experiments

the GPIB bus in order to respond with the required data. These
sometimes complex commands can be written out once in Delphi in
an interface unit specific to that instrument. Each Delphi program
that is then written uses these interface units to communicate with
instruments. A standard interface unit (called gpib) is also required
in order for the Delphi program to communicate with the GPIB
card in the computer. All the steps described in the this section need
only be performed once. After this is all setup it provides a simple
and rapid method to write new programs.
As an example consider reading the X output from a SR830 lock-
in amplifier. A simple interface unit would be as follows below. The
unit name (DCSTAN830) is given at the top, then there are two main
sections “interface” and “implementation”. The interface section con-
tains definitions of class properties, global variables and all procedures
contained in the unit. The first line of the implementation section
contains a list of other units from which code is referenced rather
than duplicated. The code at the very end encapsulated with a “begin”
and “end“ statement creates the class and sets a text handle which
corresponds to the text handle linked to the correct GPIB address
set earlier in the CBCONF32 program. There are two procedures
also in the implementation section, the first initialises the instrument
and the second “GetVolts” sends a message over the GPIB interface
(“SNAP?1,2,3,4”) and waits for the response to come back before
interpreting it and saving part of it as the variable x. These messages
and the form of the response are detailed for each instrument in its
manual.

unit DCSTAN830;

interface

type TLIA= Class

ieeehdl:integer;
ieeestr:String;
x:real;
Procedure Initialise;Procedure GetVolts;end;

var LIA:array[1..4] of TLIA;

implementation
uses dieeecb,dmisc,sysutils,gpib;

Procedure TLIA.Initialise;begin
 Making Measurements • 99

ieeeHdl:=InitDevice(ieeeStr);
ibclr(ieeeHdl);
writeieee(ieeeHdl,’OUTX1’);
writeieee(ieeeHdl,’*CLS’);
end;

Procedure TLIA.GetVolts;var s:string;begin

writeieee(ieeeHdl,’SNAP?1,2,3,4’);
readieee(ieeeHdl,s);
x:=s2r(copy(s,1,pos(‘,’,s)-1));
end;

begin LIA[1]:=TLIA.Create;
LIA[1].ieeeStr:=’SR830’;
end.

As well as the interface unit the actual program is needed - the sim-
ple one shown below sets a label on the Delphi form to show the
lock-in X value when a button is pressed. At the top in the “uses” sec-
tion the name of the interface unit (DCSTAN830) has been added:
this automatically includes the code from the file above to save it
being typed out again. The “FormCreate” function is run automati-
cally when the program is opened and runs the instrument initialisa-
tion procedure. The “Button1Click” function is run when the button
on the form is clicked. This first runs the “GetVolts” procedure to
get the value from the instrument and then sets the label caption to
be the value returned and saved to the variable x.

unit Unit1;

interface

uses
Windows, Messages, SysUtils, Variants, Classes,
Graphics,
Controls, Forms,
Dialogs, StdCtrls, dcstan830, dmisc;

type
TForm1 = class(TForm)
Button1: TButton;
Label1: TLabel;
procedure Button1Click(Sender: TObject);
private
Private declarations
100 • Physics Lab Experiments

public
Public declarations
end;

var
Form1: TForm1;

implementation
$R *.dfm

procedure TForm1.Button1Click(Sender: TObject);

begin
lia[1].getvolts;
label1.caption:=r2s(lia[1].x,13,-1);
end;

procedure TForm1.FormCreate(Sender: TObject);

begin
Lia[1].initialise;
end;
end.

This example can be extended to include all the instrument

functions that are required. In practice the programs are run on a
timed loop which reads the value on the instrument approximately
every one second and records this to a data file or plots it to a live
graph.
In order, save the data to a file it needs to be assigned a handle
using Assign File(handlename,filename); It is then opened in append
mode to add a line to the file or rewrite mode to over-write the current
contents. append(handlename); or rewrite(handlename); The line is
then written using writeln(handlename, “Text”,_tb,VariableName);
where each element is separated with a comma, text is written in
quote marks, variable values are written by including the variable
name and a tab is written with “_tb”. The file is then closed using
closefile(handlename); The “handlename” variable needs to be
declared as a textfile and the “filename” variable declared as a string.
MATLAB
MATLAB can be used in a similar way to record data and plot
graphs. The Instrument Control Toolbox must be installed. It comes
with a comprehensive manual explaining how to communicate and
control instruments via GPIB and Serial interfaces.
 Making Measurements • 101

As a brief illustrative example, the following code snippet shows

one of a number of methods data can be captured from an instru-
ment connected via a GPIB interface. Here the data is stored in the
variable “data1”.
% Create a GPIB object.
obj1 = instrfind(‘Type’, ‘gpib’, ‘BoardIndex’, 7,
‘PrimaryAddress’, 10, ‘Tag’, “);
%Create the GPIB object if it does not exist
%otherwise use the object that was found.
if isempty(obj1)
obj1 = gpib(‘ni’, 7, 10);
else
fclose(obj1);
obj1 = obj1(1)
end

% Connect to instrument object, obj1. fopen(obj1);

% Communicating with instrument object, obj1. data1
= query(obj1, ’SNAP?1,2,3,4’);
% Disconnect from instrument object, obj1.
fclose(obj1);
%Clean up all objects.
delete(obj1);

LabVIEW
LabVIEW software gives a graphical point and click interface
which allows acquisition and processing of data and data logging. It
also has the ability to make decisions based on the measurements
and control equipment. LabVIEW programs can be run on Win-
dows, Mac and Linux.
A LabVIEW program is called a VI. When LabVIEW is started,
from the “Create Project’ window select the “Blank VI” template:
see Figure 3.23. This brings up two windows: one shows the “block
diagram” where the code is developed and the other shows the
“front panel” which is where the user interface can be customized to
include graphs and buttons: see Figure 3.24.
In the “block diagram” window, pictogram components can
be wired together to create a program. Each block has inputs and
outputs just like a function in a traditional programming language.
When a component executes it produces data which passes down
the wire to the next block. The movement of data determines the
order in which the components are executed in the program.
102 • Physics Lab Experiments

Figure 3.23: Creating a new VI in LabVIEW.

The component from which the measurement is to be made

needs to be connected to the computer. A common way is via GPIB
interface or via a USB system such as a National Instruments com-
pact DAQ or an equivalent system made by Stanford Research Sys-
tems or another supplier. For example, to setup and read from the
GPIB bus, right click in the “block diagram” window and find the
“express” palette, then choose “input” and the “Instr Assist” button.
Place the “Instrument I/O Assistant” onto the block diagram. Right
click on it and choose the “Instrument I/O Palette”, the “GPIB” and
finally the “Read” button. See Figure 3.25. Place the “GPIB Read”
component onto the block diagram. Hovering the mouse over the
different input and output markers around the edge shows that one
is the “address string”. Right click on this and choose “String Pal-
ette” and then “String Constant”. Place this on the block diagram
and type in the address string for the instrument. The address string
must be set to the text handle linked to the correct GPIB address via
the CBCONF32 program as discussed in Section 3.26.2. Also set a
 Making Measurements • 103

Figure 3.24: The LabVIEW program. On the left is the ‘block diagram’ window and on the
right is the ‘front panel’ window.

numerical value which indicates the number of bytes which must be

read from the instrument.
Then wire the “String Constant” component to the “address
string” input on the “GPIB Read” component. The output from
the “GPIB Read” component is a string, this must be converted to
a number for processing. Right click on the “data” output marker

Figure 3.25: Inserting a ‘GPIB Read’ component in LabVIEW.

104 • Physics Lab Experiments

of the “GPIB Read” component and choose “String Palette”, then

“String/Number Conversion” and finally “Fract/Exp String to Num-
ber”. Place this onto the block diagram and wire it up.
To set up a chart, right click in the “front panel” window to
bring up the “controls” palette choose “Graph Indicators” and then
“Waveform Graph”. Place this on the “front panel” window. Now
right click on the input to the “Waveform Chart” component and
select the “Array Palette” option and then the “Build Array” com-
ponent. Add this to the block diagram. Now wire the output of the
“Fract/Exp String to Number” component into the “Build Array”
component and finally into the “Waveform Graph”. See Figure 3.26.
This will only produce a single value. Inserting a “while”
loop will run the program continuously until a “stop” button is
pressed. Right click in the block diagram window. Choose “Exec
Control” and then a “While Loop”. Use the mouse to draw a box
around all the components that are required to be inside the
while loop. A stop button has also appeared on the “front panel”
window. See Figure 3.27. The program can now be run, by press-
ing the run arrow in the menu bar at the top of either the “block
diagram” or “front panel” windows. Live data should appear on
the graph.

Figure 3.26: A LabVIEW program which reads data and produces a graph.
 Making Measurements • 105

LabVIEW programs can be more complex than this if more

instruments are used and the data is written to a file using the
“Write Measurement File” component from the “Express, Output”
palette. Programs for use by many users would normally be com-
piled as an executable so that changes to the code can’t be made
by mistake.

3.26.3 STILL CAMERAS AND SCANNERS

CCD sensors (or digital cameras) are a useful way to capture

data in am image format such as from diffraction/interference pat-
terns, emission/absorption spectra and with the correct optics from a
microscope. The CCD sensors used may be sensitive to a wide range
of wavelengths of electromagnetic radiation such as X-rays, visible or
infrared. A computer can then be used to make detailed measure-
ments from the images using special software.
Often the image on the computer screen will need calibrat-
ing so that a certain on screen distance (or number of pixels) can
be matched up with a certain physical distance in the real world.
Typically a ruler or graticule (a very finely spaced series of lines
used under an optical microscope) is photographed under the same

Figure 3.27: A LabVIEW program within a ‘while’ loop.

106 • Physics Lab Experiments

conditions of magnification and focus as the sample. A special soft-

ware program can be used to calculate a scale and to make mea-
surements from the image of the sample using the image of the
graticule.
Alternatively, for a small number of measurements, a simple
image editing program can be used to measure the number of pixels
between two graticule lines. Figure 3.28 shows a photograph of a
graticule taken down a microscope. Each small division corrections
to 100 µm. If this was measured to be 40 pixels, then each pixel
would correspond to 100/40 = 2.5 µm. Distances on the image of the
sample could then calculated by measuring the appropriate number
of pixels and multiplying by 2.5 µm.
A similar procedure can be followed if images are scanned into
a computer. This is perhaps most likely with images taken using tra-
ditional film and developed on photographic paper.

3.27 VIDEO CAMERAS

Video cameras, camcorders, and slow motion (high frame rate)

video cameras can be used to record experimental data. They are par-

Figure 3.28: Photograph of a graticule using a microscope.

 Making Measurements • 107

ticularly useful for measuring the speed of objects which are moving
or events which happen too rapidly to be seen with the human eye.
The experiment should be setup so that a ruler or scale is in camera
shot and in close proximity to the object. Alternatively a resolution cal-
ibration must be performed giving a value in units of meters per pixel.
Captured video can be opened in a program such as Apple
QuickTime. The video should be played back frame by frame. The
position of the object can be read from the ruler or by measuring
its position in terms of pixels from a fixed point. When the same
measurements are taken from the next frame the velocity can be
calculated using:

change in distance (m)

v (3.25)
change in time
or
change in pixels (px)  resolution (m/pixel) (3.26)
v
change in time

The change in time is obtainable from the software or by calculating

one over the number of frames per second.
If using QuickTime, once the video file is opened, choose “Show
Movie Inspector” from the “Window” menu. This window has,
amongst other details, the “Current Time”. The video can be moved
one frame at a time using the right or left arrow keys on the keyboard.

3.28 OTHER RESEARCHER’S DATA

It is frequently very useful to be able to take graphical data pub-

lished in textbooks or other authors and to be able to replot this in a
different way or in conjunction with your data. Research papers only
rarely provide tables with the actual data points which make up the
graphs, so more creative means must be adopted to extract the data.
A very useful program called Scan It [10] allows a scanned image
of a graph to be loaded. Firstly, set the axes by clicking on three well
defined points such as the origin, a point on the x axis and a point on
108 • Physics Lab Experiments

Figure 3.29: Using ScanIt to set the axis.

the y axis. For each point input the coordinates based on the scale
printed on the graph. This is shown in Figure 3.29.
The program allows the user to either identify points manually
or to automatically detect and trace a curve. Both methods produce
a table of x and y coordinates which can be exported to a graph draw-
ing program.

3.29 WII CONTROLLERS

Over recent year there has been great interest in using Nintendo
Wii games console controllers as data loggers. Wii controllers have
a built in 3 axis accelerometer and an infrared camera which can be
used to give the distance from a pair of infrared sources. They can
communicate with computers via their built in bluetooth radio.
Early work [11],[12] involved a significant amount of computer
programming. Dedicated and easy to use “Wiimote” software has
now been written [13],[14] to allow the connection of Wii controller
with a computer and subsequent collection of data in real time.
The computer must have a Bluetooth radio adapter. Sometimes
these are built into computers (especially laptops), but a cheap
 Making Measurements • 109

USB dongle can also be purchased. The more stable and reliable
ones tend to be those which use the Microsoft Bluetooth Stack
drivers. Once the computer is setup, the Wii controller needs to
be paired by pressing buttons 1 and 2 together and initiating the
pairing via the computer. If a passcode is requested in Windows
it can be left blank or skipped by pressing Alt-S on the computer
keyboard.
Once the Wii controller is paired, the Wiimote software can be
started. In accelerometer mode, the acceleration can be recorded
in the x, y and z axis. It measures in units of g between ± 3g with a
resolution of 0.04 g. When the controller is stationary lying flat on
the desk, one of the axes will measure 1 g. This can be used to mea-
sure the acceleration of a dynamics cart, a pendulum, a mass spring
system, a lift, a car or a rotating disk.
The software can also be set in “position sensor” mode. The bar
which usually sits on top of the TV is commonly referred to as the
“Sensor Bar”. In fact, it does not sense anything, but is two infrared
LEDs positioned 20cm apart. These two sources emit infrared light
which is detected by the 1024 pixel by 768 pixel infrared camera
in the Wii controller. Given the angular fields of view of the cam-
era are 41° horizontally and 31° vertically it is possible to find the
average angle subtended per pixel as 0.040°. The software in the
Wii is able to measure the number of pixels, n, between the two
light sources on its infrared camera and convert this into the angu-
lar separation. Since the sources are known to be 20 cm apart, the
distance, d, from the WiiMote to the Sensor Bar can be calculated
using triangulation:

0.1
d (3.27)
tan(0.020  n)

3.30 TRADITIONAL PHOTOGRAPHS

Photographic film may still be used to take images from electron

microscopes, X-ray cameras, or to make holograms. Where even pos-
sible, CCDs for these operations are still expensive and sometimes
110 • Physics Lab Experiments

difficult to fit to older equipment which is more likely to be present

in undergraduate laboratories. Thus it is useful to know how to
develop film and then to produce paper photographs.
The film is usually a piece of plastic with a silver halide (typi-
cally a combination of AgBr and AgI) emulsion coating and a pro-
tective gelatin layer. When the film is exposed to photons of elec-
tromagnetic radiation an electron within the halide ion absorbs the
photon, gaining energy. This frees the electron so that it is able to
move within the silver halide crystal. After some time, it can become
trapped by a defect in the crystal (often a silver sulphide impurity)
known as a sensitivity speck. The electron can then combine with
a silver ion, forming a silver atom. The silver atom is an additional
impurity which traps further electrons, creating further silver atoms.
Once exposed in the experiment, individual film packets should
only be opened in a dark room. The dark room may be dimly lit with
a colored light compatible with the film (meaning the film has a low
sensitivity to that specific color of light) or in complete darkness.
Usually the solutions necessary are placed in 4 side by side trays.
The film is first put into a developer solution. The developer is
an alkali which which preferentially reacts with the silver deposits.
Since it is a reducing agent it converts further Ag+ ions into deposits
of silver. This enhances the process which was started by the free
electrons generated by the electromagnetic radiation.
If the film is left in the developer for too long, then all the silver
halide crystals will be converted into silver atoms giving a black film.
A brief wash in water can remove all the developer and prevents any
further reactions.
The film is then placed into a third solution, the fixer. This is
usually an acid solution which removes any silver halide crystals
which have not reacted. Finally the film is washed again to remove
the fixer which prevents the film turning brown due to the formation
of silver sulphide. The films should then by hung up to dry or put
through a suitable dryer.
The exact chemicals used for the developer and fixer depends on
the film. They should be regularly changed if a large number of films
are being developed. The time the film is left in the developer and
 Making Measurements • 111

fixer solutions has a large effect on the final image and will depend
on their strength. The correct timings will usually be given, however
as a guide consider around 1 minute in the developer, 30 seconds in
the first wash, 2 minutes in the fixer, and 2 minutes in the final wash.
To create a paper photograph a second stage of development is
required. The developed film is inserted into a slot below the light
source on an enlarger. The light shining through the film creates an
image on the easel. Once the size and position of the image have
been adjusted the light on the enlarger is turned off. A sheet of pho-
tographic paper is placed onto the easel. The light source can then
be turned on using a timer, exposing the photographic paper to the
image. The final stage is to develop the photographic paper in a simi-
lar way to the development of the film. The developer solution is
usually specific to the paper, but the same two wash baths and fixer
bath as for the negative can usually be used.
Always carefully mop up any chemicals which are spilt and clear
away all chemicals at the end of each session/day as they can leave
residues behind when they evaporate. Chemicals should not be
returned to the original bottle as this will cause contamination.
 CHAPTER

4
DATA ANALYSIS
AND ERRORS

Perhaps at school, once a measurement was completed and a

result calculated it was thought the experiment was complete. Some-
times the result may have even been compared with an accepted
value. For instance, an experiment might have been performed to
measure the acceleration due to gravity, g. Perhaps the oscillations
of a simple pendulum were timed and the result substituted into the
familiar equation:

l
T  2p (4.1)
g

At university level, experiments must be taken further to include an

estimate of the error in the result. Scientists are usually interested in
making measurements, not for their own sake, but to test a theory.
Theories can be accepted or rejected on the basis of experimental
measurements, therefore it is vital to have some indication of the
accuracy of the results.
Every time a measurement is made there will be a degree of
uncertainty in the result: an instrument may be difficult to read, it
may fluctuate, it may be badly calibrated, a genuine mistake may
be made reading the instrument or the result could be limited by
the precision of the instrument. It is vital that a competent scientist
114 • Physics Lab Experiments

has an understanding of the terminology of errors as well as how

to calculate, evaluate and develop techniques to minimize the sig-
nificance of measurement error in their work. It is beneficial to
adopt an inquisitive mind-set where questions are always asked and
improvements to the apparatus and techniques which may reduce
measurement error are always considered: this can be summarized
as “be awkward”!
Suppose a large underground cavern is to be found. The hypoth-
esis might be that due to the void in the ground the measured value
of g near the cavern will be less than 9.81 m/s2. Of course, the famil-
iar value of g=9.81 m/s2 is not exact and so also has an error which is
neglected in this illustrative example.
A value of 9.75 m/s2 is measured. Does this provide evidence of
a discrepancy which may indicate the presence of the underground
cavern? To find out an indication of the error in this measurement
is needed. The result may be found to be 9.75 ± 0.01 m/s2. This is
inconsistent with with 9.81 m/s2 and so providing there is confidence
that the experiment was well performed and a good estimate of the
error made, there is evidence for the presence of the underground
cavern.
Suppose the result is found to be 9.75 ± 0.10 m/s2. This time the
magnitude of the error is a factor of 10 times larger. This means that
the result is, in fact, consistent with 9.81 m/s2 and that the result is in
agreement with what is expected. The search for the cavern should
continue elsewhere.
Finally, suppose the result is found to be 9.75 ± 5 m/s2. This
result is also consistent with 9.81 m/s2, however the error is quite
large and means that even quite significant difference can not
be detected. This raises an important point: it may not actually
be possible to detect the cavern at all if the error in the results
is too large. Frequently in the history of science, discoveries
have been made and theories accepted or rejected by scientists
who have developed their experiment and equipment to have a
smaller error than anybody had previously. If you want to make
a discovery, as emphasized earlier in this chapter you should “be
awkward”!
 Data Analysis and Errors • 115

This chapter begins with an explanation of the terminology of

errors. It then develops data and error analysis techniques with a
particular focus on graphs.

4.1 ACCURACY AND PRECISION

Figure 4.1: Targets illustrating the differences between the terms accurate and precise.
The center of the target corresponds to the true value.
The stars are the individual measurements.
116 • Physics Lab Experiments

The accuracy of a measurement is how far the measurement

is away from the true value of the quantity that is being measured.
The precision of a measurement is the full width of the spread of
data measured when taking repeat measurements under identical
conditions.
A measurement can be accurate and not precise, precise and
not accurate, precise and accurate or neither precise nor accurate.
Often either a target board (see Figure 4.1) or a distribution graph
(see Figure 4.2) is used to illustrate this.
For example a measurement may have a large systematic error.
This would mean that it was not accurate as the measurements would
lie a long way from the true value. However, the measurement could
be very precise as it could be repeated many times yielding the same
result. Thus the results may be closely scattered but around the
wrong value. A crude example would be measuring the length of a
desk to the nearest centimeter using a meter rule on which the first
5 cm was missing. Alternatively a measurement may be taken with
an instrument which has large fluctuations in the readings. This
would mean that the measurement may be accurate, but not precise.

Figure 4.2: Graphs illustrating the differences between the terms accurate and precise.
The vertical line corresponds to the true value. The vertical axis is a probability distribution
(or the number of measurements) and horizontal axis is the magnitude of
the quantity being measured.
 Data Analysis and Errors • 117

4.2 MEASUREMENT ERRORS

Measurement errors can be split up and considered as having

one of two distinct origins: they are either random errors or sys-
tematic errors. As will be seen, this distinction is important when
considering the significance of a new result or when trying to reduce
the measurement error.

4.2.1 RANDOM ERROR

Random error, as the name suggests, is inherently unpredict-

able. It causes the results to be scattered around the true value. It
arises from: natural fluctuations in readings or the environment;
or the experimenters” interpretation of the readings. The mean of
the fluctuations around the true value of a quantity has an expec-
tation value of zero. Random error is linked to the term precision
which is discussed above: a large random error indicates an impre-
cise result, conversely a small random error indicates a precise
result.

4.2.2 SYSTEMATIC ERROR

Systematic error is an error caused by the experimental tech-

nique or apparatus. It causes to the mean of the measurements to be
different from the true value. It is important to estimate the size of
systematic error as well as its direction. A systematic error may cause
a result to be larger than the true value or less than the true value.

4.2.3 ZERO ERROR

Zero error is a special name given to a common type of sys-

tematic error. This occurs when a measurement device does not
read zero before being put into use. For example an analogue
118 • Physics Lab Experiments

voltmeter may already read 1 V before being connected across

a battery to measure its potential difference. Alternatively, a bal-
ance may read 3 g before a mass is placed upon it. It is relatively
simple to account for zero errors, by making the relevant correc-
tion to each result. For example, subtracting 1 V or 3 g from each
measurement.

4.3 UNCERTAINTY

Uncertainty is the degree to which a result is unknown. In the

case of direct measurements of a quantity the uncertainty is half the
resolution. It should be quoted as a plus or minus value on the given
measurement. For example: the circumference of a can of beer
could be 18.0 cm ± 0.5 cm. In more complex cases involving detailed
analysis and non linear dependencies the plus error and minus error
might be different. For example: the current best estimate of the
mass of the up quark is 2.3+ 0.7 −0.5 MeV [15].

4.4 RESOLUTION

The resolution of an instrument is the smallest change that the

device can detect in the quantity it is measuring. It is the smallest
scale division on the meter. For example a meter rule with markings
only every 1 cm has a resolution of 1 cm. Whereas a meter rule with
markings every 1 mm has a resolution of 1 mm. A digital stop watch
giving a reading to two decimal places of a second (i.e. 1.22 s) has a
resolution of 0.01 s.

4.5 TOLERANCE

The term tolerance is usually used to express the uncertainty

that the design allows for during manufacture of an object or compo-
nent. For example a piece of metal might be needed that is 1.000 m
 Data Analysis and Errors • 119

long with a tolerance of 1 mm. This means that the metal needs to be
cut somewhere between 0.999 m and 1.001 m. Another frequently
occurring example is a resistor. One of the color bands on the resis-
tor corresponds to the tolerance. For example a 100  resistor with
a gold band (which indicates a tolerance of ± 5 %) should have a
resistance between 95  and 105 .

4.6 SENSITIVITY

In situations (e.g., using an oscilloscope) where a quantity (e.g.,

potential difference) is not displayed or measured with the same
dimensions as the measurement (e.g., cm rather than V) it is appro-
priate to define the sensitivity of a measuring device. This is the
ratio of the change of output to the change of input. A high sensi-
tivity would mean a large distance in cm corresponding to a small
(electric) potential difference whereas a low sensitivity would mean
a small change in distance corresponding to a large (electric) poten-
tial difference. In this example, the sensitivity could be given as
2 V/cm meaning the oscilloscope screen gives a reading where 1 cm
corresponds to 2 V.

4.7 RESPONSE TIME

The response time is the time taken for the measurement device
to respond and reach its final reading after a dramatic change in
input. Digital meters typically have a faster response time than an
analogue meter with a pointer because the analogue meter needs a
comparatively long time to physically move. Changes in the quan-
tity being measured which happen faster than the response time
of the measurement device will be averaged out and not seen. A
good demonstration of this would be using an alternating current
from a signal generator as the input to a center-zero analog meter.
Starting from a low frequency (with period much longer than the
response time) where the pointer is seen to oscillate the frequency
can be gradually increased until the period is much shorter than the
120 • Physics Lab Experiments

response time where the pointer will simply twitch slightly around a
reading of zero.

4.8 THE MEAN OF THE SAMPLE

Frequently in Physics measurements are repeated over and

over again. It is probably easy to recall school teachers impress-
ing the need to always repeat measurements. Why did they do
this? Repeated measurements which agree or are consistent with
each other increase the confidence that a precise result has been
achieved.
When repeated measurements are made of a quantity subject
to random error a mean can be calculated. If there are N readings
(i.e. y1,y2...yN) the mean can be found by adding up all the results
and dividing the total by the number of measurements made. This is
written mathematically as:

m y 
 N
i 1 yi
(4.2)
N

This assumes that any error of a particular size is equally likely to be

positive as negative.
For example, suppose that a measurement of the current flow-
ing through a lamp is measured 10 times and the values were:

0.563A, 0.497A, 0.501A, 0.489A, 0.512A,

0.492A, 0.536A, 0.472A, 0.520A, 0.484A
then using Equation 4.2, y 0.507. Taking the final result as the
mean of a large number of readings rather than any one individual
reading is useful as it reduces the effect of random errors. However
it does not give the whole picture. It does not give any indication of
the spread of the values around the mean. The smaller the spread
the more likely the mean is to be equal (or very close) to the true
value, providing there is no systematic error. A small spread means a
high degree of confidence can be placed in the result.
 Data Analysis and Errors • 121

4.9 VARIANCE AND STANDARD DEVIATION OF

THE SAMPLE

As a way to quantify the spread in a set of measurements the

variance can be calculated. The variance, s2, of a set of N measure-
ments of the quantity y is given by:
N

 y
1
– y
2
s2  i (4.3)
N
i 1

where y– is the mean of the set of N measurements given by Equation

4.2. s2 will often be called the variance of the sample and s the stan-
dard deviation of the sample. The standard deviation is the square
root of the variance. A large variance indicates a large spread in the
measurements and visa versa for a small variance. As an example
take the data shown in Table 4.1.

Table 4.1: Example data for calculation of the mean and variance.
122 • Physics Lab Experiments

Equation 4.3 might look over complex as the deviation of a

single measurement from the mean is given by  y i – y  and that it


1
 yi – y  .
N
would be possible to find the variance from s 2 
N i 1


 y i – y  is
N
However, this does not work in practice because i 1

always exactly equal to zero as a result of the definition of the mean.

It is valuable, especially when writing computer code, to con-
sider that Equation 4.3 for the variance can be rewritten:
N

y
1
s 
2 2
i – 2 yy i  y 2 (4.4)
N
i 1
N
x i2 2 yy i y 2
 N –
i 1
N

N
(4.5)

y 2N (4.6)
 y 2 – 2y 2 
N
 y2 – y2 (4.7)

where
N
yi2
y 2

i 1
N (4.8)

and
N

N
yi
y (4.9)
i 1

This has significant advantages when computing the variance of

a large data set on a computer as only one pass over the data is
2
required. Both y and y can be calculated in one pass, rather than
requiring a first pass to evaluate y and then a second to evaluate s2
using Equation 4.3.
 Data Analysis and Errors • 123

4.10 MEASUREMENT DISTRIBUTIONS

Figure 4.2 shows two number lines with values marked upon
them: one has a small spread and the other a much larger spread.
Both have the same mean, however more confidence could be
placed in the calculated mean being equal to (or very close) to the
true value if the data followed the pattern with the smaller spread:
this reinforces the point that the mean does not give us the whole
picture when analyzing results.
As an alternative to the number line, a set of measurements may
be represented by a histogram. Figure 4.4 shows a histogram for
the 10 measurements of the current flowing through a lamp given

Figure 4.3: Number lines comparing the spread of data.

124 • Physics Lab Experiments

Figure 4.4: Histogram of the measurements of the current flowing through a bulb given in
the text.

Figure 4.5: Histogram of the measurements of a quantity when the number of measure-
ments tends to infinity.
 Data Analysis and Errors • 125

above. As an increasing number of measurements are taken the

intervals (or bin widths) can be made smaller. A curved line (called a
distribution curve) can be superimposed on the histogram. The dis-
tribution curve passes through the middle of the top of each bar. As
the number of measurements tends to infinity the distribution curve
becomes smooth and the mean value, y , tends to the true value, Y:
see Figure 4.5. The distribution function, p(y), shown in Figure 4.5
gives the fraction of the total measurements in each interval. Thus
p(y)dy is the fraction of the total number of measurements which lie
in the interval y to y+dy. The statement is equivalent to stating that
p(y)dy is the probability that when a single measurement is chosen
at random it lies in the interval y to y+dy. Extending this interval to
cover all y values from − to + means the interval must contain our
measured value with probability 1, thus:


–
p  y  dy  1 (4.10)

and the mean of the distribution, < y > , is given by:


 y   –
yp  y  dy (4.11)

The distribution shown in Figure 4.5 is only obtained under “per-

fect” circumstances. An infinite number of independent measure-
ments would need to be taken. Clearly, in practice this is impossible
and this graph can’t be obtained from a finite number of measure-
ments. The mean value, < y > , of the distribution shown in Figure
4.5 is considered to be the true value, Y, of the quantity being mea-
sured. The actual set of measurements (for example, the 10 mea-
surements of current from above) is a random sample taken from
the distribution, p(y). The sample will have a mean value of y (as
discussed above) which differs from the true value Y by a random
and unknown amount. The point of error analysis is to estimate how
much the mean of the sample differs from the true value, that is, to
find y – Y in this notation.
126 • Physics Lab Experiments

4.11 STANDARD ERROR ON A SINGLE

MEASUREMENT

A single measurement is part of a distribution of many single

measurements like that shown in Figure 4.5. The standard error on
a single measurement, , is the half width of this distribution. When
a single measurement is made it will lie somewhere in this distribu-
tion. If the spread in Figure 4.5 is large, then the error in the single
measurement will potentially be large and so  will be large. Con-
versely, if a the spread in Figure 4.5 is small, then the error in the
single measurement will be small and so  will be small.
Technically, the standard error on the single measurement is
given by:


  y – Y  p  y  dy
2
s2  (4.12)
–

In practical terms, using a small set of real data the best estimate of
standard error on a single measurement is given by the equation:

 y
1 N
– y  s
2
s i (4.13)
N –1 N –1
i 1

where there are N different measurements labeled yi …yN. The

standard error on a single measurement does not decrease as the
number of measurements increases.

4.12 STANDARD ERROR ON THE MEAN

If the errors in the individual measurements are entirely ran-

dom, then the standard error on the mean value of the N measure-
ments may be found using
 Data Analysis and Errors • 127

s s
sm   (4.14)
N N –1

As the number of measurements taken increases, the standard

error on the mean decreases. It follows a square root relationship
so increasingly more measurements need to be taken to reduce the
error by a small amount.
It is important to understand the difference between the stan-
dard error on a single measurement, , and the standard error in the
mean, m. The standard error on a single measurement is a mea-
sure of the distribution of the measured values. This distribution
does not become narrower when the number of individual measure-
ments increases: it becomes better defined and the curve becomes
smoother. The standard error on the mean does decrease as the
number of measurements increases: the mean becomes more accu-
rately known.
Returning to the earlier example in Table 4.1 the following can
now be calculated:

6
s  9.74  10.7 (4.15)
5
10.7
sm   4.4 (4.16)
6
x  250  4 (4.17)

4.13 ESTIMATION OF ERRORS

It often occurs that repeated measurements of a simple quantity

like the length of a pendulum gives very little meaningful spread in
the results or even the same results each time. In this case calcu-
lation of the standard error on a single reading and the standard
error in the mean, as described above, is rather meaningless: the
accuracy of the measurement is limited by the resolution of the
ruler. For example consider using a meter rule with a resolution
of 1 cm: that is it has markings every one centimeter. Each time
the length is measured the scale is read to the nearest centime-
ter, so 35.6 would be read as 36. This has introduced a systematic
128 • Physics Lab Experiments

error for this individual reading, but the error in a group of read-
ings is random because it is equally likely to be positive as negative.
The error will range from nothing to half a division (0.5 cm), and so
the average error will be about a quarter of a division (0.25 cm). The
standard error (corresponding to 1 standard deviation if we assume
a normal distribution) will be about a third of a division (0.33 cm).
Care must be taken to give a reasonable estimate of the error in
measurements like this.

4.14 GAUSSIAN ERROR DISTRIBUTION

So far estimated values for the the size of errors in measure-

ments have been found. However, nothing has been mentioned
about the probability of seeing these errors. Assumptions need to
be made that the error is equally likely to be positive as negative and
that it follows a Gaussian or Normal distribution.
There is no inherent reason why this should be the case, except
that in many cases it is found to be true to a good approximation.
However, a common discrepancy is that larger errors are more com-
mon than would be suggested.
The probability that a certain reading lies between ±y0 of the
true value is

1 y0  y2 
P  y
s 2p – y0
exp  – 2  dy
 2s 
(4.18)

where  is the standard error on a single measurement.

Since the integral is difficult to solve analytically, recall the “z”
statistic which is looked up in a table of normal distribution values.

y–m
z (4.19)
s

The “z” value for data being within one standard deviation of the
mean is 1, since the difference y−µ = . This corresponds to a prob-
ability of 0.68 or 68 % that a given measurement is within 1 standard
 Data Analysis and Errors • 129

Table 4.2: Probability that a given measurement is within ‘z’ standard deviations of the
mean.

deviation of the mean. Table 4.2 shows the probabilities for higher
numbers of standard deviations.

4.15 COMBINATION OR PROPAGATION OF

ERRORS

If measurements of two or more different parameters are made

in an experiment and then used to calculate another quantity the
error in the value for this quantity must be calculated. This can be
found by using the errors in the measurements of each of the param-
eters. The simplest case to consider is when there is a linear com-
bination of parameters. More complex situations involve products/
quotients or the general case of any function.
130 • Physics Lab Experiments

4.15.1 LINEAR

Consider the equation:

a b  c (4.20)

where b and c are the measured parameters. Differentiating gives:

da  db  dc (4.21)

where a is the error in a single given measurement of a, similarly

for db and dc. In any given measurement the exact size size of the
error is not known: that is, the values of db and dc are not known.
If they were, the errors could simply added or subtracted from the
measured value to find the exact values of b, c, and then a.
The mean square deviation of a set of measurements from the
mean is given by 2 giving the mean square error in b as sb and the
2

mean square error in c as sc . These equations apply equally to  the

2

standard error on a single measurement and m the standard error

on the mean of a set of measurements, although for printing clarity
the  version will be used.
If it can be assumed that these measurements of b and c are
drawn from a Gaussian distribution, then the maximum error should
be infinite since the Gaussian distribution does not come to an abrupt
end a certain distance from the mean. A more realistic notion of the
maximum error is found by directly adding the errors in b and c giv-
ing a = b+c. However, this doesn’t give the generally accepted
rule for combining errors. It may be that the errors instead subtract
a = b−c giving a smaller result. Taking a compromise of these two
cases involves adding b and c as though they are perpendicular
and form two shorter sides on a right angle triangle. Therefore add-
ing in quadrature (i.e., applying Pythagoras’ Theorem) gives

sa2  sb2  sc2 (4.22)

 Data Analysis and Errors • 131

This can be formally proven:

sa2    a – a  
2
(4.23)
   b  c  –  b  c   
2
(4.24)

   b – b    c – c   
2
(4.25)

   b –b      c – c    2  b –b  c – c  
2 2
(4.26)
 sb2  sc2  2   b – b   c – c   (4.27)

The final term depends on whether the errors in b and c are corre-
lated in some way. If there is no correlation then the size of the error
in b has no effect on the size of the error in c and visa versa. This
means that the term  b – b   c – c  will average to zero. This reduces
the error in a2 to the form given in Equation 4.22.

4.15.2 PRODUCTS AND QUOTIENTS

Consider the equation:

a b x c y (4.28)

where the powers x and y can be any rational or irrational number

including positive values, negative values, and fractional values. Tak-
ing the logarithm of both sides and the differentiating gives

log a  x log b  y log c (4.29)

da db dc
x y (4.30)
a b c

Following the method applied in the linear case above, take the
mean square values added in quadrature:
2 2 2
 sa  2  sb  2  sc 
 a   x  b   y  c  (4.31)

A proof of this follows the same pattern as the formal proof given in
the linear case.
132 • Physics Lab Experiments

It is interesting to note that in the product case the fractional

error in the parameter is found rather than then absolute error as in
the linear case.
In the common cases of a=bc or a=b/c Equation 4.31 simplifies
to:
2 2 2
 sa   sb   sc 
 a    b    c  (4.32)

Again, Equations 4.31 and 4.32 are only valid if the errors in b and
c are uncorrelated.
It can be demonstrated that the overall error in not the same for
correlated and uncorrelated errors by considering the measurement
of the diameter, d, of a wire and then using this to calculate the cross
sectional area, A, of the wire. Taking the equation for the area as:

pd 2
A (4.33)
4

Equation 4.31 gives:

2 2 2
 sA  2  sd   sd 
 A   2  d   4  d  (4.34)

However if the equation for the area is taken as:

pd  d
A (4.35)
4

That is, we associate b and c from Equation 4.31 separately with d

then the errors b and c are certainly correlated (obviously, they are
the same). Equation 4.31 now gives:
2 2 2 2
 sA   sd   sd   sd 
 A    d    d   2  d  (4.36)

This is only half the error found in Equation 4.34 and does not give
a valid estimate of the error in A.
 Data Analysis and Errors • 133

4.15.3 GENERAL FUNCTIONS

Sometimes the analysis of an experiment depends on functions

which are more exotic than simple addition, subtraction, division,
and multiplication such as trigonometric, exponential or logarithmic
functions.
Suppose the function f is defined in terms of n different mea-
sured quantities xi:

f  f  y 1, y 2 , y 3,, y n  (4.37)

The value f0 should be calculated by setting all the measurements

yi to their measured or mean values. Next, the n different values of
fi can be found in which each parameter yi in turn is set equal to its
measured (or mean) value plus its error. Finally f is calculated using:

 f – f0 
2
sf2  i (4.38)

This has the effect of combining in quadrature all the individual

changes in f caused by moving each variable by its error.
As an example consider the function:

f  tan  x  (4.39)

If x is measured and found to be 89 ±0.5° then:

f 0  tan(89)  57 (4.40)
f 1  tan(89.5)  115 (4.41)

Now calculate:

 f – f 0    57 – 115  3364
2 2
sf2  i (4.42)
sf  58 (4.43)

This gives us an upper limit on the value of f. A lower limit can be

found in a similar way using x=88.5. Giving f = 19. The final result is:
134 • Physics Lab Experiments

f  57–19
58
(4.44)

Notice that a different upper and lower limit has been found and it
is entirely appropriate to quote the error like this.

4.15.4 APPROPRIATE SIGNIFICANT FIGURES IN

FINAL ANSWERS AND ERRORS

The number of significant figures given in a final answer should

never be more than the least number of significant figures recorded
in any of the measurements taken to find the answer.
As an example consider using the equation R=V/I to find the
resistance of a wire. The voltmeter reads 3.45 V and the ammeter
reads 0.12 A, this means the (electric) potential difference is given
to 3 significant figures, however the current is only given to 2 signifi-
cant figures. The answer should be given to two significant figures,
so R= 29 .
To consider the error in this value for resistance, let us take the
uncertainty in the ammeter reading to be 0.005 A and the uncer-
tainty in the voltmeter reading to be 0.005 V. This indicates a maxi-
mum value of R=30.04  and a minimum value of R=27.56 . It is
not appropriate nor meaningful to state that R=29 ±1.44  as the
last two digits given make no changes to the value of 29 . Errors
and uncertainty ranges should usually be quoted to the same num-
ber of decimal places as the given answer. In this case a correct state-
ment would be R=29 ±1 .

4.16 PLOTTING APPROPRIATE GRAPHS

Graphs allow trends to be spotted in data and straight line

graphs allow measurements of physical quantities to be made from
gradients or intercepts and help with spotting potential systematic
errors. It is usually beneficial to plot data on a graph as soon as to is
recorded, rather than waiting until the end of the experiment. This
 Data Analysis and Errors • 135

means that anomalous data can be spotted quickly and repeated so

that large amounts of time are not wasted.
Numbers in standard form can be entered into a computer
spreadsheet by replacing, for example, 6.63×1034 with “6.63e34”.

4.16.1 DRAWING LINES OF BEST FIT AND

FINDING GRADIENTS

A line of best fit may be drawn by hand onto a graph. The aim
should be to draw a single smooth line without using repeated short
strokes of the pencil. There should be no sharp kinks in the graph
unless the data obviously indicates this. The line need not go directly
through any of the points and should have approximately the same
number of points sitting above the line as below. Sometimes a flex-
ible ruler is useful for drawing curved lines or a ruler with a slit in the
middle (available from http://www.bestfitlineruler.com) is useful for
straight lines. The gradient of a straight line should be taken using as
large a triangle as reasonably practical. For a curved line, the gradi-
ent may be found at individual points by taking the gradient of the
tangent to the curve at this point.
Computer programs can be used to draw a line of best fit (some-
times called a trendline) based on a least squares fitting (described
later in more detail) to a mathematical function such as y=mx+c or
y=a+bx+cx2.

4.16.2 ERROR BARS ON GRAPHS

Error bars should be added to graphs for all x and y values where
possible. They give a measure of the uncertainty in the position of
the point on the axes. If a number of repeats are taken for each point
then the error bar should extend above the point by a distance corre-
sponding to m and also below the point by a distance corresponding
to m. Horizontal error bars are more usually related to the resolution
136 • Physics Lab Experiments

of the of the equipment, rather than measurement error, but never

the less should be included on graphs in a similar way.

4.16.3 STRAIGHT LINE GRAPHS

A straight line graph follows the form:

y  mx  c (4.45)

where m is the gradient and c is the y axis intercept. If an equation

can be rearranged into this format and compared with y=mx+c then
the values of m and c from the graph can be used to find the match-
ing parameters. For example, consider the EMF, Œ, of an chemical
cell of internal resistance r where values of the (electric) potential
difference V and the current I have been measured:

V  – Ir (4.46)
y  c  mx (4.47)

If the (electric) potential difference is plotted on the y axis and the

current on the x axis, the gradient will be −r and the y axis intercept
will be Œ.
Alternatively, consider the time period T of a pendulum of
length l:

l
T  2p (4.48)
g

Rearranging this into a more convenient format gives:

4p 2
T2 l (4.49)
g
y  mx (4.50)
 Data Analysis and Errors • 137

If the square of the time period is plotted on the y axis and the length
is plotted on the x axis, the gradient of the graph should equal 42/g
and the acceleration due to gravity may be measured. The y axis
intercept in this case should be zero. If a significant y axis intercept
is found then a systematic error should be suspected in the data.

4.16.4 LOG GRAPHS

Log graphs are usually used because of one, or both, of two main
benefits. Using a logarithmic scale on one or both axes of a graph
allows data which span a very large range to be plotted in a more
meaningful way. Measurements in the range 1 to 10 take up the
same proportion of the axis as measurements in the range 10 to 100
or 100 to 1000. If a log scale is used for the x axis data then logx
rather than x must be plotted on the x axis. It is most convenient to
plot a log graph using a computer although it may be done on paper
using special logarithmic graph paper (which has non-equal spacing
between the grid lines).
A second benefit is that by plotting both the x and y values on a
log scale a power law in the data can found, since:

y  xn (4.51)
log y  n log x (4.52)

If logy is plotted on the y axis and logx is plotted on the x axis then
the data should form a straight line with gradient n.

4.17 GRAPH DRAWING SOFTWARE

Files of experimental data recorded by computers are usually

in a format known as “tab delimited” text meaning the table of data
is arranged with one row per line and there is a tab separating each
column of data. Alternative delimiters include spaces, commas, or
full stops.
138 • Physics Lab Experiments

Most programs allow tab delimited text files to be opened. For

example in Excel choose “File” and then “Open”. Near the “File
name” box, there is a “Files of type” drop down. Use this to choose
the “Text files” option. Double clicking on the file (which must have
the extension .txt) starts the “Text Import Wizard” which allows the
starting row to be specified and the delimiter to be set. In Kaleida-
graph when any text file is opened the “Text File Input Wizard”
begins automatically.
It is sometimes necessary to differentiate a data set. Recall the
definition of differentiation:

df  x  f  x  d x  – f  x 
 (4.53)
dx dx

To differentiate a table of data take the difference in y values between

the current data point and the next. Divide this by the difference in x
values. For instance if x values are in column A of a spreadsheet and
y values in column B, the differential is (B2-B1)/(A2-A1) or more
generally (B(N+1)-B(N))/(A(N+1)-A(N)).

4.17.1 EXCEL 2003 AND PRIOR VERSIONS

Graphs may be plotted using a spreadsheet program such as

Microsoft Excel. Care must be taken so that a scatter graph is plot-
ted with x values on the x axis and y values on y axis. It should not be
taken for granted that Excel will plot data as you want it.
With Excel 2003 highlight data with the mouse (holding the Ctrl
button while using the mouse allows selection of non-contiguous
regions). Next, select the “Insert” menu and then the “Chart” option.
The “Chat Wizard” is displayed: see Figure 4.6. In most cases the
“XY scatter” option will need to be selected as this will plot one set
of points verses the other.
In step 2 of the “Chart Wizard” the “Series” tab allows the data
plotted on the x and y axis to be changed. This allows the axes to
be swapped should Excel plot them the wrong way round. This is
shown in Figure 4.7.
 Data Analysis and Errors • 139

Figure 4.6: Plotting an ‘XY Scatter’ graph with Excel 2003.

Step 3 of the “Chart wizard” allows a chart title and axis titles to
be set and gives a choice which grid lines to display (along with a few
other less note-worthy options).
A trend line may be added to the graph by right clicking the
mouse on one of the data points and selecting “Add Trendline”:
see Figure 4.8. An appropriate trend line should be selected (often

Figure 4.7: Selecting the data series with Excel 2003.

140 • Physics Lab Experiments

Figure 4.8: Adding a polynomial trend line of order 2 to the data and ensuring the equation
is displayed on the chart.

linear) and in the “Options” tab, the “Display Equation on Chart”

checkbox should be checked.
Error bars can be added to the graph by right clicking on one of
the data points and selecting “Format Data Series” from the popup
menu. Clicking on the “X Error Bars” tab gives the option to have
error bars display in both directions (or just the positive or just the
negative directions): see Figure 4.9. “Fixed Value” error bars can
also be chosen (this is generally appropriate where repeated mea-
surements give the same result and the estimated error is based on
the resolution), “Percentage” and “Custom” (which allows the selec-
tion of a column from the spreadsheet with the same number of
rows as the number of x values). Similar options are available for the
y axis on the “Y Error Bars” tab.

4.17.2 EXCEL 2010

With Excel 2010 highlight the data with the mouse (holding the
Ctrl button while using the mouse allows selection of non-contig-
uous regions). Next, select the Insert ribbon and then the “Scat-
ter” option as shown in Figure 4.10. In most cases the “XY Scatter”
option will needed as this will plot one set of points verses the other.
 Data Analysis and Errors • 141

Figure 4.9: Adding X axis error bars. They have a fixed size of ± 0.25.

This will insert a chart onto the spreadsheet. If Excel has plot-
ted the axes the wrong way round these can swaped by selecting the
“Select Data” button from the “Chart Tools, Design” ribbon. This
opens a window which allows the problem to be solved: see Figure
4.11.
To add titles to the x and y axes select the “Axis Titles” button
from the “Chart Tools, Layout” ribbon as shown in Figure 4.12. The
chart title can be set in the same way.

Figure 4.10: Plotting a graph with Excel 2010

142 • Physics Lab Experiments

Figure 4.11: Selecting the data series with Excel 2010.

A trend line may be added to the graph by right clicking the

mouse on one of the data points and selecting “Add Trendline”. An
appropriate trend line should be selected (often linear) and in the
“Trendline Options” tab, the “Display equation on chart” checkbox
should be checked: see Figure 4.13.

Figure 4.12: Adding axis titles with Excel 2010.

 Data Analysis and Errors • 143

Figure 4.13: Adding a linear trend line to the data and ensuring the equation is displayed on
the chart.

Error bars may be added to the graph be choosing the “Layout”

ribbon under “Chart Tools”. Then choose the small arrow to right
of the “Error Bars” button and select “More Error Bar Options”.
This produces a similar menu as Excel 2003 where there is the
choice to have error bars displayed in both directions (or just the
positive or just the negative directions): see Figure 4.14. There is a
choice between “Fixed Value” error bars (this is generally appropri-
ate where repeated measurements give the same result and there is

Figure 4.14: Adding X axis error bars. They have a fixed size of ± 1.
144 • Physics Lab Experiments

Figure 4.15: Switching between editing X and Y error bars.

an estimate of the error based on the resolution), “Percentage” and

“Custom” (which allows the selection of a column from the spread-
sheet with the same number of rows as the number of x values).
The menu which pops up relates to either vertical or horizontal
error bars. To edit the error bars for the other direction, choose the
drop down menu in the top left of the “Layout” ribbon. Either X or
Y error bars for each series on the graph can be selected. Now go
back to choose the small arrow to right of the “Error Bars” button
and select “More Error Bar Options”: see Figure 4.15.

4.17.3 KALEIDAGRAPH

In Kaleidagraph, a graph can be plotted by selecting the “Gal-

lery” menu, followed by “linear” and then “scatter”: see Figure 4.16.
The next window gives a choice of which variable is plotted on which
axes: see Figure 4.17.
Axis and plot titles can be set by double clicking on the default
label.
A trend line can be plotted by selecting the “Curve Fit” menu
and then in this case the “polynomial” option: see Figure 4.18.
 Data Analysis and Errors • 145

Figure 4.16: Plotting a graph with Kaleidagraph.

By default the fit parameters box is added to the chart. This gives
the values of the coefficients m0, m1, and m2 where y=(m0)+(m1)
x+(m2)x2. As the data plotted follows a simple y=x2 format, it is no
surprise to see that m0  0, m1  0 and m2=1: see Figure 4.19.
Error bars can be added by right clicking on the graph and
choosing “Error Bars...” from the popup menu. In the “Error Bar
Variables” window insert a cross in the “X Err” box to add X error

Figure 4.17: Selecting data for a graph with Kaleidagraph.

146 • Physics Lab Experiments

Figure 4.18: Adding a Trendline with Kaleidagraph

bars. An “Error Bar Settings” window then pops up in which the

type of error bars (Percentage of Value,
Fixed Value, Standard Deviation, Standard Error, or a Data Col-
umn) can be chosen. Y error bars can be added by putting a cross
in the “Y Err” box in the “Error Bar Variables” window. See Figure
4.20.

Figure 4.19: The fit parameters box in Kaleidagraph.

 Data Analysis and Errors • 147

Figure 4.20: Adding error bars in Kaleidagraph.

A word of caution: graphs and data must be saved as separate

files: graphs have the file extension QPC and data tables have the
file extension QDA.

4.17.4 MATLAB

To draw a graph in MATLAB, first load the data points into vec-
tors within the software. A very simple graph can be drawn using the
following code which can be entered into the MATLAB editor and
saved as a .m file:
i=0:1:100;x=i;y=i.*i;plot(x,y);

This produces a graph of y=x2 for x values between 0 and 100 in

steps of 1. The x values are stored in the vector x and the y values
are stored in the vector y. The simple plot command then produces
a graph as shown in Figure 4.21.
Extending this method means more than one line can be plotted
on the graph:

i=0:1:100; x=i; y=i.*i;

j=0:2:50;
148 • Physics Lab Experiments

Figure 4.21: A simple MATLAB graph.

x2=j;
y2=200*j;

hold on plot(x,y);
plot(x2,y2);
hold off

The line styles can be changed to dashed and the colors altered to
red and blue and axis labels added by typing:

i=0:1:100; x=i; y=i.*i;

j=0:2:50;
x2=j;
y2=200*j;

hold on
plot(x,y,’-b’,’LineWidth’,2);
plot(x2,y2,’-r’,’LineWidth’,2);
xlabel(‘x title’)
ylabel(‘y title’)
 Data Analysis and Errors • 149

Figure 4.22: A MATLAB graph showing two lines and axes labels.

hold off

This produces the graph shown in Figure 4.22.

In reality data may be saved in text files. If there are five tab
delimited text files named file1.dat, file2.dat, etc. then they can be
loaded and plotted by typing:
dataFiles=[‘file1.dat’;’file2.dat’;’file3.
dat’;’file4.dat’;’file5.dat’]; .
hold on
for i = 1:length(dataFiles(:,1))
data=load(dataFiles(i,:));
plot(data(:,1),data(:,2));
end
hold off

The properties of the graphs such as the line style, line thickness,
colors, labels, and fonts can be set in the code as illustrated above,
but they can also be modified by selecting the “Tools” menu from
150 • Physics Lab Experiments

Figure 4.23: A MATLAB graph with vertical error bars.

the graph window and choosing “Edit Plot”. Then right click on the
graph, the axes or the lines to bring up menus to allow the properties
to be modified for the individual graph.
MATLAB does not have a built in way of simply adding a best fit
line to a graph. One of the least squares method described later in
Section 4.18 must be employed. Usually the necessary computer
code is already written (often called a numerical recipe) and
the function must me called and the results it gives interpreted.
A detailed explanation of this is illustrated in Section 4.18.5.
Only vertical errors bars can be added in MATLAB. The code
needed for this is:
i=0:10:100; x=i; y=i. *i; e=0.1. *y;
errorbar(x,y,e);

which produced the graph shown in Figure 4.23.

 Data Analysis and Errors • 151

A horizontal error bar feature can be added to MATLAB by

downloading a function called HERRORBAR from the MATLAB
File Exchange.

4.18 LEAST SQUARES FITTING

For a computer the process of obtaining the best fit of exper-

imental data to a line following a particular functional form is
basically one of minimizing the sum of the square differences
between the experimental point and the calculated point for each
value of an independent variable. Efficient methods of function
minimization have long been an area of research in computer
science resulting in numerous algorithms such as the Steepest
Descent method, the Gauss-Newton method, one developed
by Levenberg and Marquardt or the Nelder-Mead [16] simplex
method.

4.18.1 REGRESSION LINE OF Y ON X

This method applies to straight line graphs and is often taught as

part of the statistics units in school math courses. The straight line is
parameterized as

y  a  bx (4.54)

where there are N data points and:

a  y – bx

  x i – x  y i – y 
N
S xy i 0
(4.55)
b 
S xx

2
 xi – x 
N
i 0 (4.56)
152 • Physics Lab Experiments

This is the method that Microsoft Excel uses to add a linear trend-
line to a graph. The R2 value given by Excel provides a measure of
the correlation between the data and the trendline. It does not, as
is often thought, provide a measure of the goodness of the fit of the
line to the data. The R value is actually the product moment correla-
tion coefficient where:

  x i – x  y i – y 
N
S xy i 0
R 
S xx S yy
 N
i 0
 xi – x 
2
  N
i 0
 yi – y 
2
 (4.57)


N
xy –
i
  x   y 
i
N
i 0
i
N
i 0
i

i 0 N


  
  
N 2 N 2

 xi  yi
 
N i 0 N i 0 (4.58)
 x i2 –  y i2 –
i 0 N i 0 N
  
  

Table 4.3: Example data for a regression line

 Data Analysis and Errors • 153

A value of R = 1 suggests there is perfect correlation between

the data and the trendline. A value of R = -1 suggests that there is
perfect negative correlation (as x increases, y decreases). A value
of R = 0 suggests there is no correlation. It should also be noted
that correlation in no way suggests that there is causation. For exam-
ple there may be a correlation between the number of cars driving
across a road junction in Cape Town and the sales at a well known
London department store, but there is no suggestion that drivers in
Cape Town are making their way to London!
Consider plotting the data points (0,4.9), (5,5.1) and (10,4.9) which
are scattered about the horizontal line y = 5 and fitting a linear trend-
line, observe that R2  0. This is despite there being a near perfect fit
between the trendline and the data. This confirms that the R2 value is
not a measure of the goodness of the fit, but one of correlation.
Take the example of the data shown in Table 4.3. The result of
the calculations is given in Table 4.4.
Substituting these into the Equation 4.57 gives R = -0.868 so R2
= 0.753. Equations 4.55 and 4.56 can then be used to find values for
a = 91.1 and b = -0.682 and finally the equation of the line of best fit:

y  91.1 – 0.682 x (4.59)

Table 4.4: Example calculations for a regression line

154 • Physics Lab Experiments

Figure 4.24: An Excel trendline, showing the equation of the straightline and the R2 value
are given by Equations 84, 85, and 86.

This is precisely what Excel finds when it plots a trendline for this
data as shown in Figure 4.24.

4.18.2 STEEPEST DESCENT

This is a relatively simple method based on following the direc-

tion in which the gradient is steepest. As an example take a func-
tion such as f(x,y)=2x4−x3+y2+6. Find the vector —f (x, y), this points
uphill, so take – —f (x, y), which must point downhill. Recall that



N
 eˆ i (4.60)
i 1 x i

which is

  
 xˆ  yˆ  zˆ
x y z (4.61)
 Data Analysis and Errors • 155

is three dimensional Cartesian coordinates. So in this example

 f  x, y   8x 3 – 3x 2  xˆ   2 y  yˆ (4.62)

Taking the starting point as (a,b), the point (a,b) must be found such
that a= a – —f(a, b) and bb – — f (a,b).  is the step size, if this
is small enough then f(a,b) > f(a,b) and so by finding f(a,b) a smaller
value for the function f(x, y) has been found. This procedure can be
reproduced iteratively until, on a given iteration f (a, b) and f(a, b)
differ by less than a set tolerance. Often the function f(a, b) would be
the sum of the square differences between a set of experimental data
points and a set of data points derived from a equation of a line of best
fit. The steepest descent method is simple, easy to apply computation-
ally and each step is very fast. It is also quite efficient at finding the
minimum point if just one exists. If more than one minimum point
exists, or if there is a local minimum, this method may struggle or reach
different answers depending upon the starting point. The downside of
this algorithm is that the progress towards the minimum gets slower
and slower and often a very large number of steps are required. Also,
this only works where the function is differentiable. If this differen-
tiation needs to applied numerically, the algorithm can become quite
slow. In the case of fitting a function to experimental data the function
f(x,y) would be the sum of the square differences between the experi-
mental data and the data generated by the fitting function.

4.18.3 GAUSS-NEWTON

Assume there are M variables in the function r(b) so that the

initial parameter vector is b0=(b1,...,bM). The function r(b) is the dif-
ference between the experimental value and the fit value given by
the fitting function.

ri  yi – f  xi, b  (4.63)
156 • Physics Lab Experiments

If there are N experimental data points there will be N different

functions r(b). As long as N  M the Gauss-Newton method will
iteratively find the minimum of the sum of the square differences:
N
S(b)   r (b)
i 1
i
2
(4.64)

To begin the calculation define:

 r1 r1 
(b)  (b)
 b1 bM 
 
J(b)       (4.65)
 rN rN 
 b (b)  b (b)
1 M

So the iterative formula used in the Gauss-Newton method is:

 
–1
b k 1  b k – J  b k  J  b k  J bk  r  bk 
T T
(4.66)

where bk is the k-th modified version of the parameter vector.

Let’s try an example. Suppose there is some experimental data:

 x1, y1    1,2.1663 (4.67)

 x 2, y 2    2,2.6726 (4.68)
 x 3, y 3    5, 5.0180 (4.69)
 x 4, y 4    7,7.6373 (4.70)
 x 5, y 5    9,11.6236 (4.71)

and it is to be fitted to the equation:

y i  b1e b2x i (4.72)

Firstly set up the functions r(b) where b=(b1,b2):

 Data Analysis and Errors • 157

r1  b1,b2   2.1663 – b1e b2 (4.73)

(4.74)
r2  b1,b2   2.6726 – b1e 2b2
(4.75)
r3  b1,b2   5.0180 – b1e 5b2
(4.76)
r4  b1,b2   7.6373 – b1e 7b2
Next calculate: r5  b1,b2   11.6236 – b1e 9b2 (4.77)

 – e b2 – b1 e b2 
 2b2 
 –e – 2b1 e 2b2 
J (b)  J  b1,b2    – e 5b2 – 5b1 e 5b2  (4.78)

 – e 7b2 – 7b1 e 7b2 
 
 – e 9b2 – 9b1 e 9b2 

Taking a starting “guess” of b0=(b1,b2)0=(1.8,0.2):

r1(b0)  – 0.0322 (4.79)
r2 (b0)  – 0.0127 (4.80)
r3(b0)  0.1251 (4.81)
r4 (b0)  0.3379 (4.82)
r5(b0)  0.7342 (4.83)

and then:

 –1.2214 –2.1985 
 –1.4918 –5.3706 
 
J (b0)   –2.7183 –24.4645 (4.84)
 –4.0552 –51.0955
 
 –6.0497 –98.0043

So:

 
–1 (4.85)
b1  b0 – J  b0  J  b0  J  b0  r  b0 
T T

 b0 –  0.0467, – 0.0104  (4.86)

  1.8467, 0.1896  (4.87)
158 • Physics Lab Experiments

The process is then repeated to find b2, b3 ... bz where there are
z iterations. If z is large enough it will eventually converge to
(1.800,0.200).
The Gauss-Newton algorithm may converge slowly or not at all
if the initial guess, b0, is far from the minimum and it can only be
used to minimize the sum of the squares differences of a function.
However, it can converge very quickly and no second derivatives
need to be calculated.

4.18.4 LEVENBERG-MARQUARDT METHOD

The Levenberg-Marquardt algorithm is more robust than the

Steepest Descent and the Gauss-Newton methods as it can find
solutions even if the starting “guess” is a long way from the solution.
If there is only one minimum, a solution (or convergence) is guaran-
teed no matter what the starting “guess”. It uses a damping factor, ,
which is adjusted each iteration. If the reduction in the sum of the
square differences, S(b) is large a smaller value for  may be used on
the next iteration, bringing the algorithm closer to the Gauss–New-
ton method. But, if there is little change is the value of S(b) then 
can be increased on the next iteration giving similar behavior to the
Steepest Descent method following the direction of the gradient.
Similar to the Guass-Newton method:
N
S(b)   r (b)
i 1
i
2
(4.88)

must be minimized where ri is the difference between the experi-

mental value and the fit value given by the fitting function. If there
are N experimental data points there will be N different functions
r(b). In each iteration step, the parameter vector b is replaced by
an improved estimate b+d. To find the vector d, take the first order
terms from a Taylor expansion:

f  xi, b + d   f  xi, b   J i  (4.89)

 Data Analysis and Errors • 159

where Ji is the same as in the Gauss-Newton method:

 r1 r1 
(b)  (b)
 b1 bM 
 
J(b)       (4.90)
 rN rN 
 b (b)  b (b)
1 M

So:
N

 y – f  x , b   J  
2
S (b  )  i i i (4.91)
i 1

when written in terms of vectors:

2
S (b  )  y – f (x, b) – J  (4.92)

Consider that, at the minimum of the sum of the square differences

S(b), the gradient of the sum of the square differences with respect
to d is zero. If this was not the case, then it would not be at a mini-
mum. Therefore:

J T J d  J T  y – f (x, b) (4.93)

which can be solved for d by simply taking the inverse of JTJ and
applying it to both sides.
Levenberg modified this slightly to introduce the damping term
mentioned earlier:

 JT J  lI    JT  y – f (x, b) (4.94)

where I is the identity matrix. If  is large, then JTJ+I  I and the

calculation of the inverse of JTJ+I is not used.
Marquardt’s modification was to replace the identity matrix with
a matrix consisting of the diagonal components of JTJ which has the
160 • Physics Lab Experiments

effect of scaling each component of d by the curvature so that bigger

steps are taken when the gradient is smaller avoiding slow conver-
gence. Thus the Levenberg-Marquardt algorithm is:

J T

J  ldiag  J T J  d  J T  y – f (x, b) (4.95)

The algorithm proceeds iteratively until any one of a number of con-

ditions are met. These are often related to the magnitude of d being
lower than a certain threshold, the magnitude of [y−f(x,b)] becom-
ing smaller than a certain threshold, a maximum number of itera-
tions being completed or the value of JT[y−f(x,b)] being lower than
a certain threshold.

4.18.5 NELDER-MEAD METHOD

As a demonstration consider a function with two parameters so

the simplex is a triangle. The function to be minimized, S, is first
evaluated at each of the three vertices of the triangle and the vertices
ordered B(est), G(ood), W(orst) with increasing values of S. From W
to B and W to G the value of the function decreases so it’s feasible
that S is smaller at values away from W so the triangle is reflected
about the line joining B and G giving a new vertex R. The point M
is defined as the point halfway between B and G (see Figure 4.25a).
Consider the case where S(R) < S(G). If S(B) < S(R) then W is
replaced with R, otherwise the simplex is extended. The line join-
ing M and R is extended to point E. Now, if S(E) < S(B) then W is
replaced with E, otherwise W is replaced with R (see Figure 4.25b).
The process now repeats by calculating a new R point until the func-
tion falls below a pre-set tolerance or is evaluated a certain number
of times.
The alternative situation is S(R) > S(G). If S(R) < S(W) then W
is replaced with R, otherwise the simplex is contracted. The points
C1 and C2 are calculated (see Figure 4.25c) and the point with the
smaller function value is called C. If S(C) < S(W) then W is replaced
with C, otherwise the point S is calculated as halfway between B and
 Data Analysis and Errors • 161

Figure 4.25: The different triangles involved in a two parameter version of the Nelder-Mead
function minimization algorithm.

W and the simplex is shrunk to BMS (see Figure 4.25d). This process
repeats until a predefined convergence criterion is reached such as
S becoming less than a certain value, failing to change between suc-
cessive iterations or a maximum number of iterations is reached.
MATLAB implements this method using its fminsearch routine
in the optimization toolbox. The MATLAB function fminsearch can
be modified to calculate and return the covariance matrix i.e., the
errors on the fit parameters using the method outlined in the appen-
dix of [16]. The declaration at the top is modified to include an addi-
tional parameter N which is passed in the function call which gives
the total number of data points and also to return the error vector.

function [x,errorvector,fval,exitflag,output] = fmin

searchmf(funfcn,x,N,options,varargin)

This additional code was added after line 382 in the code for
fminsearch and is based on the appendix of [16].
%calculate covariance matrix
%qmatrix
temp=size(v);
numberofparams=temp(1,1);
for i=1:numberofparams
qmatrix(i,:)=v(i,:)-v(i,1);
end

qmatrix(:,1)=[];
%bmatrix
for i=1:numberofparams
for j=1:numberofparams
if i==j
yi=fv(1,i+1);
y0=fv(1,1);
162 • Physics Lab Experiments

p0i=(v(:,1)+v(:,i+1))./2;
y0i=funfcn(p0i,varargin:);
bmatrix(i,i)=2*(yi+y0-2*y0i);
a(i)=2*y0i-(yi+3*y0)/2;
else

pij=(v(:,i+1)+v(:,j+1))./2;
yij=funfcn(pij,varargin:);
yi=fv(1,i+1);
y0=fv(1,1);
p0i=(v(:,1)+v(:,i+1))./2;
p0j=(v(:,1)+v(:,j+1))./2;
y0i=funfcn(p0i,varargin:);
y0j=funfcn(p0j,varargin:);
bmatrix(i,j)=2*(yij+y0-y0i-y0j);
a(i)=2*y0i-(yi+3*y0)/2;
end
end
end

covarmatrix=qmatrix*inv(bmatrix)*qmatrix’;
ymin=y0-a*inv(bmatrix)*a’;
normalizefactor=2*ymin/(N-numberofparams);
covarmatrix=covarmatrix.*normalizefactor;
%getstandarddeviation
for i=1:numberofparams
errorvector(i)=sqrt(covarmatrix(i,i));
end

%end covariance matrix calculation

A working skeletal example of this procedure in action is shown

below so that it may be copied and modified as necessary. In order
for this to work in its current form it is necessary to complete the
modification to the file fminsearchm described above.
The 11 data points in Table 4.5 are generated using the equation
y=ax2+b where a=46 and b=3:These are saved as a tab delimited
text file data.txt in the same directory as the MATLAB script quoted
below.

function fitprm = example()

options=optimset(‘Display’,’iter’); %set options for
fminsearch
x=0:1:10; %set x points at which to simulate data
Data =load(‘data.txt’); %load data file
 Data Analysis and Errors • 163

Table 4.5: Example data for Nelder-Mead method.

expData = interp1(Data(:,1),Data(:,2),x,’spline’);
%interpolate data to values at simulation points
given by x
params(1)=40; %starting parameter
params(2)=5; %starting parameter

numberofdatapoints=length(x);
disp([‘initialguess=’num2str(params)])
estimates=fminsearch(@fit,params,numberofdatapoints,
options,x,expData);
164 • Physics Lab Experiments

disp(‘Estimates’)
disp(num2str(estimates))
fitprm = estimates;

function differenceVector = fit(params,x,expData)

disp(params);
fitData =params(1)*x.*x+params(2);
differenceVector=sum((fitData-expData).^2);

When this script is run, it performs the Nelder-Mead simplex

method. As the script is in operation the procedures it follows are
detailed in the MATLAB command window. Once it has converged
on a solution it outputs the answers a=46.0089 and b=2.4174 and
the uncertainty in the results a=0.0117 and b=0.5595. It is clear
the parameter a has been accurately identified and has a very low
uncertainty and the parameter b has been slightly less well identified
however the uncertainty is correspondingly larger.

4.18.6 DISTRIBUTION TESTING

Distribution testing is a useful way to see if measured data sup-

ports a certain conclusion. Suppose that a test is carried out to see if
the volume of a gas remains constant as it is heated. It would be more
rigorous to test whether the graph of volume against temperature is
consistent with the volume being constant rather than just testing
whether a straight line gives a fit where the gradient is zero. For exam-
ple, the volume may have a sinusoidal dependence on temperature
(for some reason!). A straight line fitted to this data would certainly
give a gradient close to zero, but the volume is certainly not constant.
Distributions can be tested with the 2 parameter:
N 2
 ycalculated – yi 
x 
2

i
 si  (4.96)

where there are N data points, yi is the experimental value, ycalculated

is the theoretical y value to compare with and si is the error in yi.
Without some idea of the magnitude of si it is impossible to test if yi
and ycalculated are consistent.
Calculate the number of degrees of freedom using:
 Data Analysis and Errors • 165

vN – p (4.97)

where p is the number of parameters allowed to vary in fitting the

data. A limit can be set, typically, 5%=0.05 as an acceptance crite-
rion. Using the values of the acceptance criterion and ν a value for χ2
can be found from tables.
For example, with ν = 1, if the χ2 value found from tables is 3.8
this suggests that there is only a 5% probability that the experimental
value for χ2 would be bigger than 3.8. If the χ2 value calculated for
the experiment is less than 3.8, then it is reasonable to state that the
theoretical and experimental distributions are consistent at the 95%
level. If the χ2 value calculated for the experiment is more than 3.8,
then it is reasonable to state that the theoretical and experimental
distributions are not consistent at the 95% level.

4.19 UNCERTAINTY IN GRADIENTS AND

INTERCEPTS OF GRAPHS

Errors in gradients and intercepts of graphs can be found fol-

lowing a simple method of drawing the lines of best fit with the
maximum and minimum possible gradients consistent with the
error bars plotted around the points. Figure 4.25 shows a set of
example data plotted with x and y error bars. The points follow the
line y=2.0x+2.0 and so the gradient is 2.0 and the y intercept it 2.0.
The maximum gradient can be found by drawing a line (by hand
or manually using the drawing tool in Excel) through the top/left
of the error bars on the point (6,14) and the bottom/right of the
error bars on the point (1,4). The line therefore goes through
the points (5.5,14) and (1.2,4): it has gradient 2.3 and y intercept
of 1.2. The minimum gradient can be found by drawing a line
through the bottom/right of the error bars on the point (6,14) and
the top/left of the error bars on the point (1,4). The line therefore
goes through the points (6.5,14) and (0.8,4): it has gradient 1.8
and y intercept of 2.6. The gradient should be given as 2.0– 00..23 or
166 • Physics Lab Experiments

Figure 4.26: A graph showing the maximum and minimum gradients consistent with the
error bars.

perhaps as 2.0±0.3 and the intercept should be given as 2.0– 00..86 or

perhaps as 2.0±0.8.
More formally, for the case of a straight line the uncertainty in the
gradient and intercept can be given (with no derivation here) as:

1  N ri2
 gradient  i 1
2
 (4.98)
 N  xi – x  N –2
2
i 1

and

1 x2

 N ri2
   N   i 1
2
intercept  2 (4.99)
  i 1
N  xi – x   N –2

where there are N data points and ri is the difference in y values

between the experimental data and the straight line for each x value.
The Microsoft Excel Regression package provides a useful
way to estimate errors in the gradient and intercept of straight
 Data Analysis and Errors • 167

Figure 4.27: The Microsoft Excel 2003 Regression Window.

line graphs. This package is not always loaded into Excel. In Excel
2003 click on “Tools” and then choose “Ad Ins”. Then select the
“Analysis ToolPak” checkbox. The option to select “Data Analysis”
is available from the “Tools” menu. Find and select the “Regres-
sion” option. This brings up the “Regression” window shown in
Figure 4.27.
Once the “Input X Range” and the “Input Y Range” have been
set and the “Output Range” selected, the calculations begin. The
output looks similar to Figure 4.28.
In Figure 4.28, the intercept and gradient are given in cells B27
and B28. Cells C27 and C28 give the standard error on these values.
In Excel 2010 click on “File” then “Options”. Next, choose
“Add-Ins” from the menu at the left. Find and click once on “Analy-
sis ToolPak” in the list of Add-Ins. Press the “Go...” button. Select
the “Analysis ToolPak” checkbox. Click on the “Data” ribbon, a new
option has appeared on the righthand side called “Data Analysis”.
Press this, and then find “Regression” from the list. The “Regres-
sion” window and the output table is the same as for Excel 2003.
For more complex graphs the uncertainty in the fitting param-
eters can be found by fitting the data to an equation using a least
squares fitting method. This is illustrated in Section 4.18.5.
168 • Physics Lab Experiments

Figure 4.28: The Microsoft Excel 2003 Regression package output

4.20 FOURIER TRANSFORM

A Fourier transform allows a periodic signal consisting of more

than one frequency to have its component frequencies and relative
amplitudes identified. This can be useful in many areas of Physics
where a periodic signal is measured. It is often implemented using a
computer algorithm called a Fast Fourier Transform (FFT).
In Excel 2003, first check that the “Data Analysis” command is
on the tools menu. If it is not present, the Analysis ToolPak must be
installed: see Section 4.19.
As an example, label five columns in Excel: “Time”, “Data”,
“FFT”, “FFT Frequency”, and “FFT Magnitude”. Next fill the
time and data columns with the results from the experiment. In
this example the time resolution is 0.01 and the data is the equation
y=sin(10t)+2sin(5t).
A sometimes awkward limitation of the FFT algorithm is that
the the number of pairs of data points must be equal to N=2k where
k is a whole number: for example k = 9, 10 or 11 gives N = 512, 1024
or 2048 points. The greater k, generally the greater the accuracy of
the results of the FFT. To fit data to this constraint either remove
some, re-sample it an appropriate number of times or pad it out with
zeros.
 Data Analysis and Errors • 169

Figure 4.29: Entering the data ranges into Fourier Analysis window.

To find the sample frequency take the total number of data

points N and divide this by the difference between the times for the
first and the last points t = tN−t1:

N
f (4.100)
t

In this example N=512, t = 5.12 and f=100.

Next, fill the “FFT Complex” column. In Excel 2003, choose
“Tools, Data Analysis, Fourier Analysis”. In Excel 2010, choose the
“Data” ribbon, then the “Data Analysis” button and the “Fourier
Analysis” option. The “Input Range” is column B, cells 2 to 513 so
enter $B2:$B513. The “Output Range” is $C2:$C513. See Figure
4.27. Column C now contains the complex number results of the
FFT. The “FFT Magnitude” in column E can now be calculated
by multiplying the absolute value of the complex number (in col-
umn C) by 2 and dividing by N. In this case N=512 so the formula
=2×512×IMABS(C2) can entered into cell E2. This can be copied
down to the remaining cells below by clicking a dragging the bot-
tom right hand corner of the cell or using the key combinations
Ctrl-Shift-End and then Ctrl-D.
170 • Physics Lab Experiments

Finally complete the “FFT frequency” in column D. The general

formula for this is p multiplied by the sampling frequency f divided
by number of data points N. Where p is an integer starting from
zero. That is: cell D2 is zero, D3 is f/M, D4 is 2f/N and D5 is 3f/N
etc. Instead of manually filling in each of the rows, automatically fill
the column using the Excel “Series” function. Set D2 as zero and set
D3 in this example as =100/512. In Excel 2003, click in cell C3 then
select “Edit, Fill, Series” to automatically fill the rest of the column.
In Excel 2010, select the whole C column then from the “Home”
ribbon select the “Fill, Series” option. In both versions, now select
“Series in Rows” and then set the “Step Value” as the result from cell
D3 which is f/N and then set the “Stop Value” as N: see Figure 4.30.
Finally plot a x-y scatter graph (with a line) of “FFT Frequency”
(x axis) and the “FFT Magnitude” (y axis) in the usual way: see Fig-
ure 4.31. Do not plot more than the first half (that is the first N/2)
the rows as the FFT spectrum begins to duplicate itself.

Figure 4.30: Filling in the FFT Frequency column

 Data Analysis and Errors • 171

Figure 4.31: Graph of FFT Frequency against FFT Magnitude.

In this example, it can be seen that there are two frequen-

cies present. The lower frequency at around 0.8 Hz has twice the
magnitude of the higher frequency at around 1.6 Hz. This corre-
sponds with the simulated data fed into the FFT from the equa-
tion y=sin(10t)+2sin(5t). The standard format of sine wave is sin(t)
where  = 2f. Thus the sin(10t) term has a frequency of 10/2 =
1.59 Hz and the 2sin(5t) term has twice the amplitude and a fre-
quency of 5/2  = 0.80 Hz.

4.20.1 WINDOW FUNCTIONS

To improve the accuracy of FFT taken of experimental data, N

could be increased and data covering more periods or a larger range
of times could be used. If the data input to the FFT is not periodic
(that is it does not contain exactly an integer number of periods of
the wave) then leakage will occur in the Fourier transform spec-
trum potentially obscuring some frequencies or adding spurious
ones: see Figure 4.32. This problem can be corrected by applying
the appropriate windowing function to the data. The original data is
multiplied by a window function which drops to zero at the limits of
the data window. A number of windowing functions have been sug-
gested each with advantages and disadvantages in relation to their
172 • Physics Lab Experiments

accuracy at preserving the correct frequency and amplitude and to

which signal type (sinusoidal or random) they are best applied.
Examples include the Blackman (best at reducing spectral leak-
age), the Flat Top (best for amplitude accuracy) and None (best for
frequency accuracy). A good all round function is the Hanning:

  2pi  
H (i)  0.5  1 – cos 
 N – 1  
(4.101)


This, or any other window function, can be applied by multiplying

the i-th y data point by H(i).

Figure 4.32: Fast Fourier transform of a sine wave with frequency 10 Hz. Left panel shows
leakage with no windowing function present. Right panel shows leakage reduced with a
windowing function.
 CHAPTER

5
PRESENTING RESULTS

This chapter looks at how results taken in the laboratory are

recorded and presented to others. Each week, or for each experi-
ment a script will be given which outlines what must be done and
what must be measured. As students progress through a degree
and gain more experience of work in the laboratory the scripts will
become less detailed, leaving more room for them to work out the
procedure themselves.
Almost certainly, results will be recorded and general notes
taken about the experiment in a laboratory notebook. Students may
be asked to produce a brief write up and complete the analysis for
each experiment and hand this in. A small number of experiments
are usually chosen and written up in more detail and students may
produce a poster or even give a presentation on a particular experi-
ment.
This section focuses in some detail on the layout and content of
a laboratory notebook and formal report. It gives an overview of how
a computer might be used to enhance a report and include special
characters, equations, tables, graphs, and references.
This section should be extremely valuable in giving an overview
of recording and presenting results and hopefully pointing students
in the correct direction for further investigation on the Internet. A
shortage of space means that it can not explain every detail of every
software program which may encountered, nor can it explain the
specific details of how a particular university requires work to be
produced.
174 • Physics Lab Experiments

5.1 LAB BOOK

While working it lab it is essential to keep a good record of every-

thing done so that it may be possible to repeat the work from only
these notes after a significant period of time has passed, perhaps a
year or two. The following advice is generally considered good prac-
tice for keeping a lab book:
● Each experiment should begin on a new page with an under-
lined title and date.
● A brief description of the method being carried out is useful
along with annotated diagrams or printed photos of the appa-
ratus which show the procedure and exactly which measure-
ments are being taken.
● All measurements should be recorded directly into the lab
book. Initially, no mental arithmetic or calculator work should
be carried out on them.
● Try to record all data in tables. Make sure a careful note is
made of the name of the quantity being measured, its units
and if possible an estimate of the error or uncertainty in the
readings.
● Try to use appropriate units and powers of 10. Do not try to
measure the work function of a metal in Joules, rather use
10−19 Joules or even eV in the table heading.
● Do not write on rough paper or copy pieces of data from one
page to a neat copy later. Try to stick to one, original, reason-
ably neat and presentable table.
● Lab books should have crossings out and corrections through-
out! If not it appears although they have been written up at a
later date. This not only wastes time but can cause copying er-
rors as well as potential loss of data written on loose sheets of
paper.
● Take care to record the full filenames and locations of any
data files stored on computer which need to be analyzed at a
later date. Often these will take the form of a name followed
by sequential numbers. For instance mydata.000, mydata.001,
 Presenting Results • 175

mydata.002, or mydata000.txt, mydata001.txt, mydata002.txt.

Remember to make a copy on a memory stick or email a copy
before leaving the lab.

5.2 REPORTS

As well as keeping a neat lab book showing all the results and
analysis of each experiment, students will often be required to write
up two or three experiments formally as though they were scien-
tific articles. This is a vital part of becoming a good Physicist, as
once a cutting edge experiment is completed, the results analyzed
and a new discovery has potentially been made the results need to
be communicated with other Physicists. This has traditionally been
achieved by writing the experiment up as a scientific paper and then
submitting it to a scientific journal. The journal then ask a num-
ber of other Physicists (the identities of whom are unknown the the
authors) working in similar fields at other universities or research
laboratories to be a referee for the paper and to make critical com-
ments upon the work. Referee comments are passed back to the
authors for their comment or so that they make modifications to
their paper. Based upon these comments the journal editors will
make a decision to accept or reject the paper for publication.
The aim of writing formal reports is to develop skills in writing
papers: these will be vital in a future career in academia and almost
certainly useful for a future career in any other field. They provide a
good opportunity to gain a deeper understanding of an experiment
beyond that achieved in a one or two day laboratory session.
This section gives an overview of the typical structure of a formal
written report and provides some advice on how to include refer-
ences and write a report on a computer.

5.2.1 SOFTWARE

Simple reports with a small number of equations, figures,

and references can be most simply written in Microsoft Word,
176 • Physics Lab Experiments

OpenOffice Writer, or an equivalent computer program. These

programs are convenient since they immediately reflect changes
made to formatting and content of the report, making it easy to
see how the report will appear when printed. However, number-
ing and referring to equations, graphs, and tables is more tedious
especially with a longer report. It can also at times be difficult
to maintain consistent formatting and spacing throughout a long
document and to insert special characters and complex equations.
Microsoft Word can be obtained as part of the Microsoft Office
suite of programs from either an Internet or high street shop.
OpenOffice Writer is free and can be obtained from the Open
Office website [17].
A further alternative to traditional software programs is to
use Google Docs which is a web based document creator and
editor. Its advantage is that the document is stored on Google’s
servers, so can easily be edited at a number of computers and
also jointly worked on by a number of people (useful for collab-
orative work). At the time of writing, Google Docs is sufficient
for more basic work, although some of the more advanced fea-
tures of Microsoft Office and OpenOffice are not yet present:
for example different error bars for each data point can not be
included on graphs.
More complex reports, perhaps those for final year projects
or PhD theses are best written in LaTeX (pronounced Latek):
in fact this book was written in LaTeX. LaTeX has a reasonably
significant entry barrier and learning curve. It is written as a
simple text document with all formatting, equations, and figures
inserted by using text based commands embedded within the text
of the document. This makes it somewhat difficult to visualize
a document without compiling it and then opening the result-
ing pdf file. The text based commands also need to be learnt or
a guide referred to. However equations, special characters and
references can easily be inserted and tracked and the layout pre-
cisely controlled. To start running LaTeX first install a TeX dis-
tribution such as MiKTeX [18]. It makes the compiling simpler if
a text editor designed for use with LaTeX such as WinEdt [19] is
used.
 Presenting Results • 177

5.2.2 STRUCTURE

Write ups of experiments should provide enough information

for a reader to understand the aims and outcomes of the experi-
ments, the approach taken, the measurements made, the analysis
conducted and the results obtained. They should be written in a for-
mal style and the language used should be precise: the level should
be such that it should be understandable by a peer who may not
have any detailed knowledge of the specific experiment. The data
and analysis should be presented in a manner that allows the reader
to evaluate the significance of the results. The use of the words “I”,
“you” and “we” should be avoided: the report should be written in an
impersonal style i.e., the third person.
A formal laboratory report should be self-contained and struc-
tured so that it is easy to follow. A common approach is to break
the report into a number of sections with clear headings. Sections
should be numbered, as should diagrams, graphs, and equations.
This allows easy and precise referral during the text of the report:
for instance to state that “using Equation 2.2 the data shown in Fig-
ure 2.3 can be used to plot the graph in Figure 2.4”. Bold fonts,
underlining, numbered lists, and bullet points can be used to make
the report easier to follow for the reader.
A common structure which could be followed in most cases:
Title
A short title which describes the experiment.
Abstract
A few sentences (perhaps less than 100 words) which summa-
rizes the experiment and the results, giving quantitative information
about the findings. The abstract should not contain references to the
rest of the paper or other work.
Introduction
The introduction sets the scene for the experiment. It should
describe the background to the problem under investigation
explaining why it is interesting and encompassing a critical view of
other work on the same problem: what has been done and what still
remains to investigate.
178 • Physics Lab Experiments

Theory
A theory section should explain the physics behind the problem
and show how the experiment carried out and the quantities mea-
sured can be related to the problem.
Experimental Methods
A description of the apparatus used, including a diagram. This
should include information on how the apparatus was built or
connected together and any special features of the equipment or
particular problems which had to be overcome. A description of
the measurements taken and any special techniques which were
required.
Results and Discussion
Raw data, especially large tables of numerical data should be
placed in an appendix at the end of the report. This section should
show how the data was analyzed to give a final result and contain any
important intermediate results. A small number of graphs or sum-
mary tables are appropriate, although most extensive data sets can
be put into an appendix. It is vital to include an estimate of uncer-
tainties and to consider all sources of experimental error in the final
results.
Conclusion
The results of the experiments, together with appropriate
uncertainties, should be summarized and compared with other
values. The results should be set in context and their significance
noted. Suggestions for possible future work can be made where
appropriate.
References
Sources of information such as other papers, articles, and books
should be listed here.
Appendix
Detailed tables, graphs, derivations, or computer programs
should be presented in the appendix.
In Microsoft Word or OpenOffice Writer, the headings
may be formatted to have a specific style: perhaps a larger font
and in bold. In LaTeX, simply use the “section” commands as
follows:
 Presenting Results • 179

\section{Section Title}
Some text.
\subsection{Subsection Title}
Some more text.

5.2.3 GRAPHS, TABLES, AND DIAGRAMS

Data should always be presented in graphs as they are efficient

at showing any trends as well as scatter in the data. Graphs should
always have titles, axis labels and units: make sure they are large
enough to be read when the graph is scaled to fit into a report.
Numerical values on axes should always increase in equal steps occu-
pying equal distances on the page/screen (except for a log graph).
However the values do not need to start from zero: for example an
axis may be labeled 50, 60, 70, 80 but not 50, 55, 60, 70, 80. Axes
should be scaled so that the data points on a graph occupy as large a
portion of the graph as reasonably possible. For example, see Figure
5.1, which shows good and bad scaling for the axes of a graph. Graphs
should always be scatter graphs. Never draw line graphs where the
points are joined like a dot-to-dot. If a computer generated fit line is
used the parameters and equation should be clearly displayed. For
example see Figure 5.2 which shows good and bad practice for add-
ing lines to graphs. Finally, show x and y error bars where possible as
illustrated in Figure 5.3.

Figure 5.1: Good and bad scaling of axes.

180 • Physics Lab Experiments

Figure 5.2: Good and bad lines on graphs.

Figure 5.3: Always add x and y error bars where possible.

 Presenting Results • 181

An important skill of report writing is appropriate selection of

data. Try to find ways to plot several sets of data in one figure. For
example if there are several sets of data of resistance against length
of a wire (perhaps with different cross sectional areas) these could
be plotted on the same axes using different shaped markers (crosses,
plus signs, triangles, and circles) or different colors (only if the report
is printed in color). This will help distinguish the data sets. A legend
should also be included which links the different markers or colors
to the relevant cross sectional area.
Tables should be displayed with clearly labeled column head-
ings, titles, and units. In large tables, it is usually essential to have
grid lines, although these may be omitted in smaller tables.
Smaller graphs, diagrams, or tables may be inserted into a larger
figure: perhaps in a region of the graph where there is no data. All
figures should have a self contained caption which explains what the
figure shows. For example: “Figure 1: (electric) potential difference
plotted against current. The filled squares are data using a carbon-
zinc cell; the open squares are data from a lithium polymer cell. The
curve is a fit to Equation 3”.
All tables, diagrams, and graphs should be discussed in the text
and be relevant to the experiment performed. They should be placed
within the text, close to the place where they are being discussed
rather than collected at the end of the document.
Microsoft Word and OpenOffice Writer
Images which contain diagrams can be inserted in Microsoft
Word 2003 by selecting the “Insert, Picture, From File” menu and
then choosing the correct image file. Tables can be copied from
Microsoft Excel or inserted using the “Table, Insert, Table” menu.
Graphs are most easily copied from Microsoft Excel. In OpenOffice
Writer, the menu choices are identical.
LaTeX
In LaTeX images can be inserted using the following com-
mands:

\begin{center}
\begin{figure}
\includegraphics [width=4in] {folder/filname.png}
182 • Physics Lab Experiments

\caption {Graph showing the change in A as a

function of B.}
\label{graphname}
\end{figure}
\end{center}

The width of the image can easily be modified by specifying a num-

ber of inches or cm. The label command gives the graph a name so
that it can be referred to automatically by its figure number else-
where in the report by using the command:
\ref{graphname}
Graphs must be inserted as images. They may be screen captured
from Excel using the Print Screen key on the keyboard. They can
then be pasted into an image editor (such as Irfanview [20]), appro-
priately cropped and saved as an PNG image file. If using Kaleida-
graph then a graph can be exported using the “File, Export, PNG”
menu option.
Tables may be inserted using the following commands:

\begin{table}
\begin{tabular}{ll}
\hline x & y=x$^2$ \\
\hline 1 & 1 \\
2 & 4 \\
3 & 9 \\
4 & 16 \\
\hline
\end{tabular}\label{tablename}
\end{table}

These commands to insert images and tables mean that LaTeX will
determine their position in the document to best suit the flow of test
and figures across pages. If insertion at a fixed point in the text of
preferred, regardless of how the text flows around, omit the begin/
end figure, or begin/end table commands.
To create bullet point lists in Latex, use the “itemize” structure:

\begin{itemize}
\item The first item
\item The second item
\item The third item
\end{itemize}
 Presenting Results • 183

and the create numbered lists use the “enumerate” structure:

\begin{enumerate}
\item The first item
\item The second item
\item The third item
\end{enumerate}

5.2.4 REFERENCING

When writing a report always make clear which parts are written
by the author and which parts are taken from, based upon, derived
from or copied from other sources. For instance it is not possible to
state “It is well known that the charge on an electron is 1.6×10−19 C”.
Statements like this should refer to a source which backs it up, for
example: “The data in the present report give a value of the charge
on an electron e=2.4±0.2×10−19 C, which compares well with the
study of Price and Arthur [18] who obtained e=2.2±0.1×10−19 C”.
Always check references carefully, especially the volume and
page numbers and the spelling of authors’ names/any unfamiliar
words in the title.
There are two accepted styles for labeling references: the
numerical system and the Harvard system.
Referencing: The Numerical System
The numerical system is generally in wider use in the sciences,
especially Physics. It is the method adopted by most journals for
the publication of original research. References are listed in a num-
bered list at the end of the paper. In the main text of the paper, the
contribution of a reference in indicated with the relevant number
either as a superscript34 or in square brackets [34]. It is good prac-
tice, although not essential, for the references to be numbered in the
order in which they are referred to in the text.
Various pieces of software can help with this and make the job of
ordering numerical references less tedious. Microsoft Word has the
option for endnotes. To insert a new reference choose “Insert, Ref-
erence, Footnote”, then choose the “Endnotes” bullet point. This
184 • Physics Lab Experiments

will give a list at the end of the document, rather than at the end of
each page (if the “Footnote” option is chosen).
In LaTeX, a separate bibliography file can be created with the
file extension “.bib”. In the “.tex” document, the bibliography file is
linked by including the following lines of code:

\bibliographystyle{unsrt}
\bibliography{file_name}

This “unsrt” style produces a list in which the references are listed
in the order in which they are referenced in the text. Changing the
style to “plain” gives a list in which the references are alphabetical by
first author surname. The “.bib” file is a text file with a list of refer-
ences. For example, the entry for one of the references used in this
publication is arranged as follows:

@Article{liquidoxygen,
title = Making Liquid Oxygen,
author = French, M. M. J. and Hibbert, Michael,
journal = Phys. Ed.,
volume = 45,
pages = 221,
year = 2010,
}

and the reference for a website is given by

@MISC{winedt,
note = http://www.winedt.com
}

The list of references is inserted into the LaTeX document with

\bibliographystyle{unsrt}
\bibliography{references}

Referencing: The Harvard System

Most books and journals which predominantly publish review
articles use the Harvard System of referencing. References are listed
alphabetically by first author surname and are labeled by the Author
surname and year of publication. In the main text of the paper, the
contribution of a reference is indicated with the relevant surname
and year e.g. [French 2012].
 Presenting Results • 185

It is reasonably easy to keep track of Harvard style references in

Word with a simple list of references at the end. In LaTeX, a “.bib”
file can be used as described in the numerical references section but
the style must be set to “alpha”. This gives a reference which includes
the authors” initials and the publication year. More flexibility with
the layout of these references in Word can be obtained by installing a
special plugin and in LaTeX by using an additional specialized pack-
age.

5.2.5 EQUATIONS

Microsoft Word
In Microsoft Word 97 - 2003, equations can be inserted using
the “Insert, Object, Microsoft Equation 3”. This brings up a tool
bar, from which Greek characters, mathematical expressions, and
layouts can be selected and inserted into an equation object. Figure
5.4 shows the tool bar being used to select a fraction arrangement.

Figure 5.4: The Equation Tool Bar in Microsoft Word 2003.

186 • Physics Lab Experiments

In Microsoft Word 2010, equations can be inserted by choosing

“Equation” from the “Insert” ribbon. See Figure 5.5. A further
“Equation Tools, Design” ribbon appears which gives a full range
of options of mathematical expressions and layouts which can be
inserted as shown in Figure 5.6. Greek characters are inserted by
returning to the “Insert” ribbon and choosing “Symbol”. Then select
“More Symbols” and in the “Symbol” window, the font should be
changed to “Symbol” as shown in Figure 5.7.
LaTeX
In LaTeX, all equations are inserted by typing a text based com-
mand. For example fractions can be inserted by typing
$\frac{a}{b}$
which produces [a/b] inline with the text or

\begin{equation}
\frac{a}{b}
\end{equation}

which produces
a
(5.1)
b

which is a numbered equation within the document. Brackets which

automatically scale to the appropriate height are produced using the
commands:

\begin{equation}
\left[ \left( x \right) \right]
\end{equation}

Figure 5.5: Inserting an Equation in Microsoft Word 2010.

 Presenting Results • 187

Figure 5.6: The Equation Tool Bar in Microsoft Word 2010.

Figure 5.7: Inserting Greek characters in Microsoft Word 2010.

188 • Physics Lab Experiments

which produces

  x (5.2)

To cover a number of other common functions consider the follow-

ing example:

\begin{equation}
\int_0^\pi A\times\cos^2 \theta \partial\theta = A\
times\int_0^\pi \left(\frac{e^{\imath\theta}+e^{-\
imath\theta}}{2}\right)^2
\partial\theta
\end{equation}

which produces
2
p p  eiq e–iq 
 A  cos qq  A   q
2
 (5.3)
0 0 2 

Other symbols can be found in Table 5.1 or looked up on the inter-

net using a search engine. For example to find how to produce a
much less than arrow search for “Latex much less than”. The inser-
tion of Greek characters and subscripts in LaTeX are described in
the next section on Special Characters.
OpenOffice Writer
In OpenOffice Writer, an equation is inserted by selecting the
“Insert, Object, Formula” menu. This behaves as a hybrid between
the text based method in LaTeX and point and click interface in
Microsoft Word. Figure 5.8 shows the equation entry environment.
Equations can either by typed into the text window at the bottom
or entered by clicking in the “Selection” box. Greek letters can be
inserted by choosing “Catalog” from the “Tools” menu.
 Presenting Results • 189

Figure 5.8: Inserting an Equation in OpenOffice Writer.

5.2.6 SPECIAL CHARACTERS

Often when writing reports or labeling tables and graphs charac-

ters will be needed which are not present on the keyboard such as a
superscript 2, a degrees symbol or a Greek character.
In some software programs it will be possible to turn on super-
script or subscript options by finding a Format Text menu. However,
this can be difficult or tedious in some applications. Much more sim-
ply a superscript 2 may be inserted by holding down the Alt key on
the keyboard and then using the number pad on the righthand side
of a standard keyboard (or by engaging the num lock function on
a laptop) to type the combination 0178 and then releasing the Alt
key. A superscript 3 may be entered by typing 0179 and the degrees
symbol by entering 0176.
190 • Physics Lab Experiments

In some software programs, it may be possible to change the

font to Symbol which allows a Greek character to be typed using
the standard characters on the keyboard. A little trial and error
will allow the selection of the correct character (although some
common ones are a=, b=, m=, q=). Capital Greek letters can
be produced by turning on Caps-Lock or holding the shift key, as
usual.
Further number codes or characters maybe found and cop-
ied on an individual basis using the “Character Map” program.
This can usually be found (well hidden!) under “Programs” or “All
Programs” in the “Start Menu” in the “Accessories, System Tools”
folder. Alternatively it can be run using the “Run” command from
the “Start Menu” (or Windows button and R) and typing “char-
map”. Characters can be selected, copied, and then pasted into an
application.
In Microsoft Word, some special characters can be inserted by
following the above instructions for inserting an equation.
In LaTeX, special characters are generally simple to insert.
Greek letters are inserted by typing, for example:
$\theta$
which produces  and a capital Greek letter is inserted by typing
$\Theta$
which produces . Superscripts are inserted by typing, for example:
$x^3$
which produces x3, but
$x^34$
produces x34, where as
$x^{34}$
produces x34. Subscripts are inserted by using the underscore char-
acter instead of the hat in a similar way.
There is a truly phenomenal array of symbols which can be
inserted in to a LaTeX document. An Internet search engine can
be used to search for a general list of “LaTeX special characters” or
a more specific character by name. A brief list of some commonly
used symbols is given in Table 5.1.
 Presenting Results • 191

Table 5.1: Commonly used LaTeX symbols

192 • Physics Lab Experiments

5.3 POSTERS

It is common at scientific conferences for all attendees to be

invited to produce a large A1 or A0 sized poster describing their
recent work. These are displayed on boards, usually in a central
location to give everyone a chance to view them and discuss the
work. At university, students might be given the chance to do this
with an experiment they have carried out during their laboratory
sessions, with a final year project or at the beginning of a PhD. It
is likely that some guidance will be given on content, layout and
length. However, a good outline would begin with a title, a list of
authors and an abstract. The number of words should be kept to
the minimum needed to give an introduction to the work, explain
the experiment(s) carried out and give some analysis of the results
and a brief conclusion. The emphasis should be on the inclusion of
diagrams of the experimental setup where appropriate and graphs
or other way of displaying the results. It is crucial to ensure that
text is printed in a font size that is large enough to read: text should
be not smaller than 28 point (perhaps 24 point for captions or ref-
erences).
There are two examples of posters included in this book. Figure
5.9 was produced at the end of a final year research project and Fig-
ure 5.10 as a summary of an experiment undertaken during the early
stages of a PhD.

5.3.1 SOFTWARE

Posters can be produced in a programs such as Microsoft Word,

OpenOffice Writer, or LaTeX, however it is more straightforward
to use Microsoft PowerPoint or perhaps OpenOffice Impress. A
single slide can be used, but the page size should be enlarged to
the appropriate size. This is preferential to enlarging a normal slide
as font sizes will retain their familiar sizes. Usually posters will be
printed by an IT or printing department at the university: large
poster printers are still expensive to buy and run and there may
only be one per department. The paper will usually be a role of a
 Presenting Results • 193

certain width. Posters may be either portrait or landscape, but to

avoid distortions when printing it is useful to find out the paper
width or maximum printing width before starting a poster. Then
either the height (landscape) or width (portrait) of the poster should
be set to this.

5.3.2 PAPER SIZES

Standard European paper sizes follow the A series with A0 being

the largest, decreasing in size to beyond A6: see Table 5.2. There are
three rules governing this series:

Table 5.2: Papers Sizes.

194 • Physics Lab Experiments

1. A0 paper has an area of 1m2

2. The length divided by the width is 2  1.4142

3. The shorter side of size A(n) is the long side of A(n+1)

(that is A(n+1) is A(n) cut in half parallel to it shorter side).
Sizes of paper larger than A0 are not defined as clearly, but can
follow the pattern in Table 5.2.
There is a less common series called the B series. The size for
the length or width is given by:

B  n   A  n   A  n – 1 (5.4)

Tolerances for the paper size are ±1.5 mm for lengths up to 150mm,
±2mm for lengths between 150 and 600 mm and ± 3mm for any
dimension above 600 mm.
In Microsoft PowerPoint 2003, the page size is changed by select-
ing “Page Size” from the “File” menu and changing the “Width” and
“Height” settings. In OpenOffice Impress, the page size is changed
by selecting “Page” from the “Format” menu and changing the
“Width” and “Height” settings.
Finally, it should be ensured that images are of sufficient resolu-
tion that they will not appear “blocky” and distorted when they are
printed out on a poster. Ideally there should be at least 150 dots per
inch and probably less than 300 dots per inch to avoid excessively
large file sizes. So if a graph is required to appear about 6 inches
wide (15 cm) on the final poster, it should be between 900 and 1800
pixels wide.

5.4 PRESENTATIONS

Research scientists also regularly give presentations/talks/

seminars on their research work. Most universities make giving pre-
sentations a compulsory part of the undergraduate degree. Topics
range from focusing on an experiment done in the laboratory, a final
year project/disertation topic or on a certain topic set as a reading
assignment.
 Presenting Results • 195

Many people are nervous about or dislike giving presentations

especially to large audiences. This is something which significantly
improves with practice: try to take opportunities to practice speaking
in front of people. If presentations are well prepared and have been
practiced to friends or in front of a mirror they become easier.

Figure 5.9: Example poster 1. Based on a final year project.

196 • Physics Lab Experiments

Common mistakes with presentations are writing a full script

and trying to read it. This means that there is constant looking down
at a script and not looking at the audience. Do not rush: speak slowly
and clearly so that everyone can understand what is being said. Try
to stand somewhere so that it is easy to engage with the audience.

Figure 5.10: Example poster 2. Based on early work towards a PhD.

 Presenting Results • 197

Too many people stand behind a table holding notes and appear to
the audience to be “hiding” or seeking protection behind furniture.
Slides should not be blocks of text, perhaps a maximum of 30
words per slide as a general guide, nor should they have more than
one graph per slide. Elaborate slide transitions and flashing text
should always be avoided. Make sure text is large enough to read
(this includes graph axis labels) and the color of the text and back-
ground aid rather than hinder its legibility. For example, black on
white is easy to read, black on dark blue or yellow on white is very
hard to read. For ease of reading fonts should not be so complex
and fussy that the audience struggle to read them, although recent
research suggests [21] that if serif fonts such as Times New Roman
or Monotype Corsiva are used, the audience will remember more of
the presentation.
Start with a title slide. This should be followed with a slide giving
a brief outline of the talk. A good way to plan a short presentation
is to complete the outline slide and then turn each point from the
outline into a slide title. Finally, include an acknowledgments and
references slide unless this information is included elsewhere. As a
rough guide around one slide per minute is needed, so a 10 minute
presentation should have 10 slides in addition to slides for the title,
outline, references, and acknowledgments.
Figure 5.11 shows an example set of slides for a short presenta-
tion. The accompanying notes are as follows:
1. Originally a Sci-Fi creation, which was first proposed scien-
tifically by Carl Sagan in 1961.
2. It hasn’t been proven that humans can live in enclosed envi-
ronments for long periods of time i.e. space stations.
3. Correct Temperature - not too hot (velocity of particles
is greater than escape velocity of planet) and not too cold
(can’t heat it enough); Ability to retain an atmosphere - size
of planet - enough gravity; Location of planet - solar winds
remove atmosphere; magnetic field existence/strength;
Within easy traveling distance of earth - transport of equip-
ment, other resources if needed, people; Availability of
resources - things needed to terraform - lots of water and
CO2; Mars - close, once had water flowing on surface,
198 • Physics Lab Experiments

probably has frozen water at north pole and CO2 at south

pole, has days unlike the moon, days of the similar length,
has seasons.
4. To start terraforming - first stage is raise the temperature
of planet by an estimated 4–20 K to melt southern cap of
CO2 and start “runaway” greenhouse effect. Methods to
do this include reducing albedo by covering polar caps
with carbon or other black material to increase absorption
of heat, using large orbiting mirrors to redirect sun light,
the Russians have already built and tested in Earth orbit a
20 m space mirror in the Znamia Project, Using SF6; CFCs
no good since they are destroyed by UV light which breaks
the C-Cl bond - Mars has no ozone layer to reduce UV light
so CFCs don’t stay around for very long - would require
rapid replacement - PFC’s have no C-Cl bond so would be
much more long-lived - very efficient since only a few ppm
would be needed to provide a significant warming, they are
non-toxic, have long life times and could easily be made on
the Martian surface. Ammonia containing asteroids - am-
monia is a potent greenhouse gas, also the rock of the as-
teroid would contain a great deal of CO2 Start of runaway
greenhouse effect - CO2 melts and is released to atmo-
sphere from south pole, CO2 also evaporates from surface
rock, planet warms up, more CO2 melts, planet warms up
more… This is a proven method to warm up a planet - it’s
what’s happening here on Earth and should take approx 100
years to reach stability. A 4°C initial temp rise would eventu-
ally produce a 55°C temp increase. Human efforts would set
the ball rolling, but nature would perform 99% of the work.
5. Atmospheric pressure increased - evaporation of CO2 in-
creases pressure; Water vapor in atmosphere - Mars needs
a great deal of water transferred into vapor in the atmo-
sphere - equivalent to a slab of ice 350 miles in diameter
and 100 yards thick, this also helps to raise the tempera-
ture; Microbes and Plants introduced to increase O2 level
- microbes, especially genetically engineered or selectively
breed have to be used first since they are very hardy, plants
wouldn’t survive in the initial atmosphere of almost all
 Presenting Results • 199

CO2, algae and plants can be introduced later when there is

slightly more O2 approx partial pressure of 20 mbar, to help
create oxygen from the CO2. The content of N2 would also
have to be increased.
6. Significant increase in atmospheric pressure - currently
Mars is 7 mbar, Earth is around 1013 mbar, with the evapo-
ration of all the CO2 in the polar caps and locked in rocks

Figure 5.11: Example presentation slides

200 • Physics Lab Experiments

on the surface and before the introduction of microbes and

plants the partial pressure of CO2 could reach around 500
mbar; Liquid water available - melting of polar caps and
water under surface due to temp and pressure rises UV and
cosmic radiation decreased - excess oxygen can form into
ozone (O3) in the atmosphere, and protect the surface from
UV radiation Danger that Mars becomes too hot and enter a
quasi-stable high temperature regime.
7. References:
http://www.users.globalnet.co.uk/ mfogg,
Space 2000, Harry L Shipman, Plenum Press (1987)
The Road to the Stars, Iain Nicolson, William Morrow and
Co (1978)
www.nasa.gov

5.4.1 SOFTWARE

To create an electronic presentation use software such as Micro-

soft Powerpoint, OpenOffice Impress, Prezi, or Google Docs.
Powerpoint and OpenOffice produce conventional presentations
with a series of slides which are stored on a computer. The Google
Docs web based software has a “Slides” application which can be
accessed online from http://docs.google.com is very useful when
working on a group presentation as it can be shared with other group
members allowing them to access and modify the presentation. It
also makes working on the presentation easier from more than one
computer. More and more universities are now switching over to
cloud based software like this. Finally, for a slightly different take
on a presentation try Prezi[22]. This produces a presentation a little
similar to a mindmap which is navigated around in a defined order.
 CHAPTER

6
COMMON
EXPERIMENTS

The set of experiments completed in the first year of an under-

graduate physics course aims to develop experimental skills and build
confidence with new apparatus. Sometimes an experiment’s main
goal is to develop ability to carefully record data and apply the analy-
sis techniques described earlier in this book. A number of universi-
ties start with the measurement of the acceleration due to gravity, g.
Whilst the value of g should be familiar to every student, the goal of
this experiment is to measure it to one part in 1000. This experiment
trains students to calculate the uncertainty and error in their work
and find ways of minimizing them. It is common for experiments to
then focus on the measurement of physical constants.
Over the course of the first year in the laboratory students will
develop the ability to carefully read and follow written instructions,
apply the data analysis techniques above to carefully recorded data,
learn to use new apparatus, research, and write formal lab reports
and hopefully enjoy experimental physics.

6.1 EXAMPLE EXPERIMENTS

This section gives two example laboratory scripts. The first

looks at the measurement of the acceleration due to gravity using a
202 • Physics Lab Experiments

pendulum and the second looks at two methods to measure the

speed of light. Some scripts, or parts of scripts will be described in
detail giving you specific guidance about the measurements to make
and the equations needed to complete the analysis. Others, which
are perhaps designed for times when students have had prior expe-
rience of similar work, will only give an outline of what should be
measured and suggest how a derivation be started to find the equa-
tions needed for the analysis.

6.1.1 Measuring Acceleration due to Gravity

This experiment is often the first which you will undertake in the
undergraduate laboratory. The aim of this experiment is to measure
the acceleration due to gravity to a precision of one part in a thou-
sand. It also has the objective to teach you to understand errors. You
will learn how to treat random error in measurements of the time
period of the pendulum and how to combine errors from different
sources to give an estimate of the error in the final value of the accel-
eration due to gravity.
You will use a rigid pendulum pivoted about a knife edge. For
small oscillations the period, T, of the pendulum is given by:

I
T  2p (6.1)
mgh

I is the moment of inertia of the pendulum about the axis of rotation

along the knife edge, m is the mass of the pendulum and h is the
distance from the pivot to the center of mass of the pendulum. The
moment of inertia is the rotational equivalent of mass. If we consider
a simple pendulum (that is, a point mass on a a thin wire) which has
I=mh2 then Equation 6.1 reduces to the more familiar T  2p h g .
Rearranging Equation 6.1 gives:

4p 2 I
g (6.2)
mhT 2
 Common Experiments • 203

In order to measure g to 1 part in 1000, you need to consider errors

in I, m, h and T. Since T appears as a square term in Equation 6.2,
you need to find T to 1 part in 2000. The errors in I, m, h are not
independent since I is related to m and h, so the error may cancel to
some extent. When considering random errors, it is important that
you don’t miss the possibility of systematic errors being introduced
by your measurement process or technique.
Finding T to 1 part in 2000
The pivot is first clamped to the desk and the rigid pendulum is
suspended from its knife edge. Care should be taken to ensure that
the pendulum is swinging properly from the knife edge so that there
is no sideways oscillation or movement.
Equation 6.1 is valid if the oscillations are small as it depends
upon the approximation of sin   which only true for small .
Therefore you should keep the amplitude of oscillation to a mini-
mum: perhaps below ± 20 mm.
The pendulum should be set in oscillation and the time period,
5T, for five complete oscillations should be measured using a stop-
watch. A marker should be used in your view below/behind the pen-
dulum to define a fixed point to start and stop the stopwatch. To
minimize the error, this should be in the middle of the swing as it
is here the pendulum is traveling the fastest. This means there will
be the smallest ambiguity in the time at which the pendulum passes
this point. Six readings xi of the period 5T should be measured. Then
calculate:


6
1. The sum of the six readings xi
i 1

2. The mean period, 5T , by dividing the sum of the periods

from step2, by six
3. The six differences  xi – 5T  between the measurements
and the mean 5T
  x – 5T 
6 2
4. The sum of the square of the differences i 1
i

5. The standard error in a single

  x – 5T 
6 2
measurement s  1 5  i 1
i
204 • Physics Lab Experiments

6. The standard error on the mean of 5T sm  s 6

7. The standard error in the period, T, by diving the standard
error in the mean of 5T by 5.

You will find that you are probably some way off achieving an error
of 1 part in 2000 or 0.0005. You could, of course, continue to repeat
the 5 period method enough times as the standard error in one
period decreases as the number of repeats, n, increases. However
due to the 1 n dependence a very large number of repeats will be
required.
An alternative approach is to reduce m by reducing  rather
than increasing n. You can do this by increasing the number of peri-
ods measured from 5: this will reduce the random error, but also the
systematic error in the period T.
If your reaction time is around ± 0.1s then to achieve 1 part in
2000, you will need to measure for around 200 seconds. To ensure
you reach the target of 1 part in 2000, try to measure a time of
around 300 – 400 seconds. However, it will be very easy to lose
count of the number of periods in such a long time so try the fol-
lowing method.
Using the stopwatch time the period of an integer number of
oscillations (it doesn’t matter exactly how many) which take around
60 seconds in total. The actual number of periods can be found by
using your previous value of T and knowing there was an integer
number of periods. You should now be able to fill in the table below:

Estimate of T, from above

Time for N swings
Implied value for N
Improved estimate for T
Current estimate of error in T
Using your value of the error in T, you should find the error in your
calculated value of N. This should be much less than 1, otherwise
you will need to need repeat this step timing for a period shorter
than 60 seconds.
 Common Experiments • 205

You should repeat the procedure, but timing for around 300 -
400 seconds and fill in the table again:

Estimate of T from timing for  60 seconds

Time for N swings
Implied value for N
Improved estimate for T
Current estimate of error in T

You should now have a value for T with an error of less than 0.0005.

Finding h and m
These should be relatively straightforward. The mass m can
be determined by placing the rigid pendulum onto a balance.
The calibration of the balance should be checked using standard
masses and a number of repeats taken. The mass and the error in
the mass should be recorded. If the pendulum has mass around
1kg, then a measurement to the nearest 10th of a gram will give an
error of one part in 10,000. To measure the height h the center of
mass of the pendulum must be found by balancing the rigid pen-
dulum on a knife edge. The pendulum rod can then be marked at
the balance point. The measurement h should be taken from the
knife edge at the top of the pendulum rod to the mark you have
just made. Over a distance of around 500 mm, you will need to
measure to the nearest 0.5 mm to achieve an error of less than 1
part in 1000.
Finding the Moment of Inertia
The final measurement to be taken is the moment of inertia,
I, of the pendulum about the knife edge. With a compound object
like the rigid pendulum, the moment of inertia can be most eas-
ily found by using the “parallel axis” theorem. This states that the
moment of inertia, I, of an object about a given axis is related to
its moment of inertia, Ip, about a parallel axis through the center
of mass by:

I  I p  mh 2 (6.3)
206 • Physics Lab Experiments

Figure 6.1: Moment of inertia of a pendulum.

where m and h have the same definitions we have already used. See
Figure 6.1.
You should find that Ip is only about a tenth of the magnitude of
mh2 and so only needs to be measured to a precision of about 1 part
in 100.
Ip, the moment of inertia of the pendulum about the axis
through its center of mass, can be found by comparing torsional
oscillations of the pendulum with those of a regularly shaped rect-
angular bar:
2
I p  Tp 

I b  Tb  (6.4)

where Ib is the moment of inertia of the bar about its center of mass,
Tp is the time period of the torsional oscillation for the pendulum
and Tb is the same for the bar.
 Common Experiments • 207

First, find Ib. This is relatively simply as it is given by a standard

formula for a regular rectangular bar:

12 
a b 
mb 2 2
Ib  (6.5)

where a and b are the length and width of the bar and mb is the mass
of the bar. Since you are looking for Ip to 1 part in 100, mb is needed
to 1 part in 100 and a and b are needed to 1 part in 200.
You should suspended the pendulum from a wire about its cen-
ter of mass. The pendulum can then be displaced so that it under-
goes a torsional oscillation. This means that it twists from side to
side in the horizontal plane, with the center mass staying directly
beneath the point of suspension. The time period of the torsional
oscillation should be measured following a similar procedure to that
used previously. This is repeated for the bar. Since you are aiming
for an error in Ip of less than 1 part in 100, the time periods must
have an error of less than 1 part in 200.
To determine the errors in Ib and then Ip do the following. Write
a +b2 = X then:
2

  X  2  a  a   2  b  b
2 2 2
(6.6)

and calculate X. You may like to try proving this using the equa-
tions given in Section 4.15 of this book. The error in Ib can now be
found from:
2 2 2
 I b   mb   X 
 I    m    X  (6.7)
b b

The error in Ip is found using:

2 2
 I p   I b   T p 
2 2
 Tb 
 I p    I b   4  T p   4  Tb  (6.8)
208 • Physics Lab Experiments

Finding g and the error in g

Since:
4p 2  I p  mh 2 
g (6.9)
mhT 2

The equation for the error in g will be somewhat cumbersome.

If Ip/Ip < 0.01 then the error may be small enough to be ignored
and a simpler equation may be derived to calculate the error in g.
For the purpose of the error calculation:

 (6.10)

so:
2 2 2
 g   h   T 
 g    h   2  T  (6.11)

If Ip/Ip > 0.01 then a more complex equation must be derived for
the error in g.
Discussion
Finally, it is worth identifying the largest sources of error in the
experiment and writing a discussion focusing on whether the error
in g met the target of being less than 1 part in 1000. You should
consider how the value of g could be improved either by modifica-
tion of the experimental technique or by redesigning the method
completely. Systematic errors also need to be considered: is your
location on earth or height above sea level likely to influence the
result significantly: by how much?

6.1.2 Measuring the Speed of Light

The aim of this experiment is to make measurements of the

speed of light by two different techniques. Firstly using a modern
method, sending pulses of light down optical fibers and measuring
the delay time with an oscilloscope and secondly using a method
 Common Experiments • 209

developed by Foucault in the 19th century. It will give you practice

at using a range of apparatus, implementing precision measurement
techniques and considering the safety risks associated with the use
of lasers.
Optical Fiber Method
In this method, you will measure the time taken for photons
of visible light to propagate down various lengths of optical fiber.
However, you can’t make a simple speed equals distance over time
calculation for one fiber as there is a time delay t0 due to the elec-
tronics generating and detecting the photons which is not known.
The total time taken for a photon to propagate from the transmitter
to the detector can be given as:

nx
t  t0  (6.12)
c

where n is the refractive index of the optical fiber, x is the length of

the optical fiber and c is the speed of light. If a series of measure-
ments of t for fibers of different lengths are performed, the gradient
of a graph of t verses x can be used to find a value for c.
To begin, connect one of the shorter cables between the light
emitting diode (LED) transmitter and the photodiode fiber optic
receiver. Make sure the LED transmitter is connected to the power
supply and turn it on. So that pulses of light are produced, connect
the TTL output from a signal generator to the input on the LED
transmitter.
Next connect the signal generator to the CH1 input of a high
frequency oscilloscope. Note that a normal oscilloscope won’t be
able to sample often enough to resolve the very short time differ-
ences you will find. Connect the fiber optic receiver to CH2 input of
the same oscilloscope.
Setup the oscilloscope so that you can see two traces show-
ing square waves on the display. The traces should be enlarged so
that you can see one or two periods of the square wave. The CH2
trace should rise at a slightly later time than the CH1 trace, corre-
sponding to the time delay due to the electronics and the length of
the optical fiber between the LED transmitter and the fiber optic
receiver.
210 • Physics Lab Experiments

You may need to adjust the position of the ends of the fiber optic
cables to optimise the amplitude of the received signal, especially
with the longer cables where the detected signal is weaker. The cur-
sor controls on the oscilloscope can be used to measure the time
delay. Choose the most appropriate points to measure between and
try to justify your choice.
You should consider how reproducible your result is and whether
or not it should be repeated. Finally, repeat the measurement with a
range of optical fibers up to 100 m in length.
To analyze these results, plot a graph (including error bars for
both quantities) of the time delay verses the length of cable. Use the
graph to determine the delay time due to the electronics and the
ratio n/c. The refractive index, n, will be given to you, allowing you to
calculate c. Using multiple lines on the graph, estimate the error in
n/c and use this to express the error in c. How many standard devia-
tions is your result away from the defined value? Consider the effect
of systematic errors: where might they arise, what could be done to
minimize them and would they lead to an increased or decreased
value for c?
Foucault’s Method
A beam of laser light is focused using Lens 1 and then passed
through a beam splitter. Part of the light enters the microscope and
part makes its way to the rotating mirror. When light is incident on
the rotating mirror it is reflected towards a fixed, spherical mirror.
The fixed mirror always reflects the light back along the same path,

Figure 6.2: Faucault method diagram.

 Common Experiments • 211

returning it to the rotating mirror and then the beam splitter. Some
will be directed towards the microscope where it can be viewed in
conjunction with the light which traveled the shorter route. A dia-
gram of the apparatus is shown in Figure 6.2.
First, lets consider the point of reflection on the fixed mirror as
the angle of the rotating mirror changes. Figure 6.3 shows the path of
a light ray which is reflected from the rotating mirror at angle 2. At
a slightly later time, the rotating mirror has moved and now the light
ray reflects at angle 2(+). This means the two light rays reflect
from the fixed mirror at different points. If the distance between the
rotating and fixed mirror is D then, the distance between the two
points of reflection on the fixed mirror is given using the general
equation for the perimeter of a circle r:

S  2D dq (6.13)

Now consider a small wave packet or pulse of laser light. If the

rotating mirror is rotating as the pulse strikes the rotating mirror

Figure 6.3: Reflection from the fixed mirror as the rotating mirror changes angle.
212 • Physics Lab Experiments

Figure 6.4: Considering the virtual image of the fixed mirror.

(which is at angle ) the pulse will reflect up to the fixed mirror

and then back to the rotating mirror. However the rotating mirror
has rotated, by angle +. If the mirror had not been rotating the
returning pulse of light would have focused at the point p in Figure
6.2. Instead, it will focus at a different point p which we must find.
It is simpler to consider the virtual image of the fixed mirror as
shown in Figure 6.4. We now must consider (the thin) Lens 2. The
virtual image of the two points of reflection on the fixed mirror are
in the focal plane of Lens 2 but separated by a distance S, with one
of the points on the principal axis. Thus, from thin lens theory we
know that an object of height S in the focal plane will be focused in
the plane of point p with a height (−v/u)S. In this equation v is the
distance between the lens and image and u is the distance between
the lens and the object. The negative sign can be ignored since we
don’t mind the image being inverted. The beam splitter directs a
fraction of the light towards the microscope, giving a similar image
with height p. Therefore:

v X
dp S S (6.14)
u YD

where X is the distance between Lens 1 and Lens 2 minus the focal
length of Lens 1 and Y is the distance between Lens 2 and the rotat-
ing mirror. Substituting from Equation 6.13 gives:

2DX dq
dp (6.15)
YD
 Common Experiments • 213

The angle  depends on the rotational speed of the rotating mirror

and the time it takes the light pulse to travel from the rotating mir-
ror, up to the fixed mirror and back again (a distance 2D). The time
taken is:
2D
t (6.16)
c

and recalling the rotation analogue of distance equals speed multi-

plied by time is  = t gives:

2D w
dq  (6.17)
c

where  is the angular velocity of the rotating mirror and c is the

speed of light. Substituting this into Equation 6.15 gives:

4XD 2w
dp (6.18)
cY  D 

Rearranging gives our final equation for c:

4XD2w
c (6.19)
d p Y  D

We derived Equation 6.19 by assuming the image was created by

a pulse of laser light at a particular position of the rotating mirror.
However, Equations 6.13 and 6.17 only depend on the difference in
the angle, that is the difference in the position of the rotating mirror
in the time it takes light to travel between the mirrors. A continuous
laser beam can be considered as a series of short pulses: the image
of each pulse will be displaced by the same amount giving a single
image. By making measurements of the displacement of the image,
the angular velocity of the mirror and the necessary distances a value
for the speed of light can be calculated.
Care should be taken with the laser used in this experiment.
Refer to Section 2.7. You should not look into the microscope when
the rotating mirror is stationary unless an attenuator is in place.
214 • Physics Lab Experiments

To begin, the equipment must be aligned. Turn on the laser.

The lenses and beam splitter should already be fixed in the correct
position. Using the edge of a piece of paper carefully follow the path
of the laser beam to the rotating mirror. The rotating mirror should
be positioned by hand so that the laser beam is reflected up to the
fixed mirror.
The fixed mirror should the be adjusted so that the reflection of
the laser beam strikes the center of the rotating mirror. After placing
the attenuator in front of the microscope, the microscope should be
focused on the image of the beam. It should appear as a small disk
shaped dot. The microscope should be moved to bring the image
into the center of the cross hairs and adjusted so that it is in focus.
The position of the microscope should be recorded using the
Vernier scale attached to it. The microscope should be moved and
then brought back to align with the image at the center of the cross
hairs. The position should be recoded again. Repeating this around
ten times allows the standard error on a single reading to be calcu-
lated.
The rotating mirror should then be turned on. It should rotate in
a clockwise direction (looking down on the equipment as in Figure
6.2). The attenuator can be safely be removed. The angular velocity
of the mirror is displayed, but sometimes takes a number of minutes
to settle to a constant value. After waiting a while, the stability can
be checked by recording  every 10 seconds for a few minutes. This
enables the calculation of a value for the standard error in .
Take the first reading by turning the rotation speed of the mirror
up to a maximum and then once the speed has stabilized, bring the
microscope image to the center of the cross-hairs. Record the rota-
tion speed of the mirror and the position of the traveling microscope
using the Vernier scale. This should be repeated for a number of
lower rotational speeds.
Measurements should also be taken with the mirror rotating coun-
terclockwise. Finally the distances X, Y, and D should be measured.
A graph should be plotted of microscope position p against
mirror rotational speed . The gradient of this graph is:

w
m (6.20)
dp
 Common Experiments • 215

The value of the gradient found can be substituted into Equation

6.19 to give a value for c. The error in a single measurement of p
and  is found by calculating the standard deviation of the sets of
10 results taken earlier. Error bars should be added to the graph of
microscope position against mirror rotational speed. These should
be used to draw a maximum and minimum possible gradient and
calculate the error in the gradient.
A suitable expression can be derived based on Equation 6.19
for the error in c. Hint, it may be useful to express Y+D as another
quantity W and find the error in W. Making this substitution into
Equation 6.19 gives:

4XD 2
c (6.21)
mW

It is interesting to consider which term contributes most to the error

and is this a single dominant term? How could the random error in
these measurements be reduced? Were there any sources of system-
atic error and can corrections be made? What effect does the beam
splitter have, how big is it, and is it significant?

6.2 EXAMPLE REPORTS

The following section gives three examples of formal labora-

tory reports, based around different experiments to the examples
already studied. The first example looks at finding the charge mass
ratio of an electron, the second looks at radioactivity and the third
looks at measuring the latent heat of vaporization of liquid nitrogen.
Although all these reports were graded as firsts, they are not repro-
duced here because they are perfect and on no account should they
be copied. With careful thought there will always be ways in which
they can be improved in terms of clarity, detail and content. How-
ever, they are provided here so the reader can see examples of the
length, depth and quality that is required in a good report at under-
graduate level.
216 • Physics Lab Experiments

6.2.1 Example Report: Charge Mass Ratio of an

Electron

Abstract
Electrons emitted from an indirectly heated cathode were accel-
erated through known electric and magnetic fields and by varying
the solenoid current, focused into a narrow beam that then hit a
fluorescent screen. When the beam was focused on the screen the
electrons performed an integer number of helix rotations. For a
series of anode–cathode potential differences, measurements of the
solenoid current and number of helix rotations were taken. After
careful analysis of the experimental data acquired, the charge:mass
ratio, [e/m], of the electron was calculated to be 2.99 × 1011 Ckg−1
±0.55 × 1011 Ckg−1.
Introduction
The calculation of the charge:mass ratio of the electron was
important in the early investigation and discovery of the structure
of the atom and in proving the existence of units of matter smaller
than the atom.
J.J. Thompson first carried out an experiment similar to this
in 1897. He was attempting to disprove the widely believed idea
that the hydrogen atom was the lightest particle in existence. His
experiment, in which he measured the deflection of charged par-
ticles in a magnetic field perpendicular to the direction of motion
of the particles, allowed him to show that the electron was nearly
2000 times lighter than the hydrogen ion. This experiment is based
on an adaptation of Thompson’s, first carried out by H. Busch in
1922.
e
This experiment aims to calculate and to test Busch’s adapta-
m
tion of Thompson’s experiment.
Theory
The equation F=Ee + Bev gives the force (measured in New-
tons) on an electron of charge e (measured in Coulombs) and mass
m (measured in kg), moving with velocity v (measured in m/s), in
a magnetic field B (measured in Tesla) and electric field E (mea-
sured in volts/meter). Since the acceleration due to this force is
 Common Experiments • 217

 E  Bv  e , neither e nor m can be found individually from this

a
m
kind of experiment as only their ratio appears in the equation of
motion. The ratio can be found from the paths taken by electrons
passing through known electric and magnetic fields. Both electric
and magnetic fields are required since the path in an electric field
e dl
alone is independent of m (In the time, dt  , that the electron
v
takes to travel a short distance, l, the direction of motion changes
dt dl
by dq  a  a 2 , where  is the acceleration transverse to the
v dv
path. If the electron moves in only an electric field,  and v2 are
e
always proportional to , therefore and the paths taken are inde-
m
e
pendent of . When there is also a magnetic field,  has a com-
m
ponent proportional to v so the path depends on (e/m)1/2, which can
then be calculated.)
The most obvious way to perform this experiment, given the
cathode ray tube, is to apply the magnetic field transverse to the
e
electron beam and to calculate from the deflection of the beam
m
across the screen. Using this method, it is very hard to achieve a good
accuracy since the position of the spot is badly defined. These prob-
lems can be avoided by using the focusing effect of a longitudinal
magnetic field on a nearly parallel beam of electrons. Since all the
electrons started from the cathode with zero velocity (kinetic ener-
gies from thermionic emissions are usually less than a few tenths of
an eV and can safely be ignored) and have been accelerated through
a potential difference (the potential difference of the anode with
respect to the cathode), as the beam of electrons emerges from the
hole in the last disk of the anode, all the electrons have the same
velocity, v, given by eV = 1/2mv2. But, since the cathode has a finite
area the directions of motion of the electrons are spread out i.e.,
the electron has a small component of velocity, vt, perpendicular to
the axis. In a uniform magnetic field parallel to the axis, the force
experienced by the electron is perpendicular to the axis and to vt and
has magnitude Bevt. This has the effect of rotating the direction of vt
but since it is always normal to it, to leave its magnitude unchanged.
218 • Physics Lab Experiments

The angular velocity of vt is equal to the acceleration perpen-

dicular to vt divided by vt i.e.

Bev t m Be
w 
vt m

 is constant for a given B. This causes the electron to travel in a

helical path towards the screen. The radius of this helix is

v t mv t
r 
w Be

and the period of the motion is T=(2m)/ Be . The period depends

upon e and m and not v, the velocity of the electron. Since the period
is the same for all electrons, B can be chosen to give an integer num-
ber, n, of helical rotations as the electrons travel between the anode
and the screen. This means that the electrons produce a sharply
focused spot on the screen - see Figure (6.5). When n is not an inte-
ger number, the spot is badly focused - see Figure (6.6).
If the anode-screen distance is L then the time taken for the
electrons to travel this distance is L/v , then

L LBe m
n 
Tv 2pm 2eV

Left: Figure (6.5) Electron paths when both have an integer number of helical rotations pro-
ducing a focused spot. Right: Figure (6.6) Electron paths when both do not have an integer
number of helical rotations producing an unfocused spot.
 Common Experiments • 219

So:

e 8p 2n 2V
 2 2
m LB
e
Where is the charge:mass ratio of the electron, n is the number
m
of helix rotations, V is the anode-cathode potential difference, L is
the anode-screen distance and B is the magnetic field.
The field inside an infinitely long solenoid is given by B=µ0NI
where µ0 is the permeability of free space, N is the number of turns
per meter and I is the current flowing in the solenoid. A correction
factor of K is needed since the solenoid is only of a finite length.
Where

X A D Z U A D W 
K ln    ln 
2D  A  Y  2 D  A  V 

and A, D, X, Y, Z, U, V, W are indicated on Figure (6.7).

Experimental Methods
This experiment was carried out by finding the various fields for
which the beam was focused. For various values of V (the anode-
cathode potential difference), readings of solenoid current were

Figure (6.7): Dimensions of a solenoid used to find the correction factor, K.

220 • Physics Lab Experiments

taken, using an ammeter, for points when the spot was in focus
(i.e., integer values of n). As the current was increased from zero,
the first time the spot was in focus was assumed to be n = 1, with n
increasing by one each time the spot was in focus.
The solenoid was measured to find its dimensions to get a value
e
for K and to calculate .
m
After B had been calculated for each current, a graph of B2
against n2 was plotted so as the gradient was

e L2
m 8p 2V

Figure (6.8): Diagram of the cathode ray tube.

 Common Experiments • 221

Figure (6.8) shows the center of the apparatus, including the anode,
cathode, and the fluorescent screen. This is surrounded by the sole-
noid.
Results
The value of K was calculated to be 0.87. L, the anode-screen
distance was measured to be 0.051 m, the total length of the sole-
noid, L0, was measured to be 0.13 m and the total number of turns
on the solenoid was 1500. This gave N as 11500 turns/meter.
e
The gradients of the graphs gave for each (electric) potential
m
difference as:
Discussion
After plotting the graphs for each of the (electric) potential dif-
ferences, their gradients were found not to agree with each other
exactly, but to vary slightly. An average value for the gradient was
taken as the final answer that was around 1.7 times the accepted

e
This gives an average value of as 2.99 ×1011 Ckg−1.
m
222 • Physics Lab Experiments

Figure (6.9): Graph of B2 versus n2.

value. This means that the gradients of the graphs I plotted from my
experiment were a little too large.
The error bars on the graph show the errors with respect to the
uncertainty in the solenoid current. The error range was obtained by
using the extremes of these error bars to calculate the gradient and
e
then . They do not show, what is probably the most significant
m
source of error in this experiment, the error with respect to deciding
the exact time when the spot was focused on the screen. This was a
visual measurement which is prone to error, which cannot be mea-
sured accurately.
Even though the gradients are a little large, the line of best fit
still passes very close, but slightly to the positive side of the origin
in all cases. This suggests that there could be some systematic error
involved in these readings - which has a greater effect at larger val-
ues of n. Supporting this idea is the fact that the values of B should
be multiples of the value for n = 1. This does not quite fit the experi-
mental data. The values fall short of what this predicts them to be–
and increasingly more so as n increases.
One possible cause of this could be the partial focusing of the
beam of electrons before it emerges from the anode. To eliminate
 Common Experiments • 223

this the deflection of the spot when the electrons acquire a large
transverse component of velocity in a known part of their path
can be measured. This transverse component may be produced
by applying a potential difference between one pair of deflector
plates. The spot comes into focus when the electrons have trav-
eled an integer number of turns in their helical path from the
center of the deflector plates to the screen. The distance L, in
the earlier equation, can be substituted for the distance between
e
the deflector plates and the screen, and calculated using the
m
same methods.
Also, using an AC (electric) potential difference between the
deflector plates could make it easier to see when the spot is in focus
since the spot is then drawn out into a line and a collapse of the line
into a spot indicates an integer number of helical rotations. Since
the points lie so close to the lines of best fit, there is little evidence
to suggest any significant random error in this experiment. As well
as the extension involving applying AC (electric) potential differ-
ences to the deflector plates, a larger CRT could be used - as this
would make it easier to see the spot and when it’s in focus/not in
focus.
Conclusions
From my experiment the charge:mass ratio of the electron is
2.99 ±0.55 ×1011 Ckg−1. Considering the inaccuracy in the visual
“measurement” of when the spot is in focus, I consider my answer to
be consistent with others” results, even though the accepted value of
1.76 ×1011 Ckg−1 is outside my error range, it is of the same order of
magnitude and only a factor of 1.4 out. I consider Busch’s adaptation
of Thompson’s experiment to valid - especially if the above improve-
ment of applying an AC (electric) potential difference to the deflec-
tor plates behaves as predicted.
References
[1] Haliday, Resnick, Krane, “Physics”, Fourth Edition, 1992,
pp738 – 740
224 • Physics Lab Experiments

Appendix
 Common Experiments • 225

6.2.2 Example Report: Investigating Radioactivity

Abstract
Some of the properties of radioactive count rate and the
Geiger Muller tube were investigated. The binomial distribution
for independent events was approximated by the Normal and Pois-
son distribution. The Poisson was found to be a better approxima-
tion since the probability of decay was small. The dead time of
the Geiger Muller tube was found to be 495 ± 73 µs. The inverse
square law for the count rate was tested with a gamma and a beta
source, the results showed an almost perfect fit to theory. Gamma
rays were found, as expected, to be more penetrating than beta
particles. Exponential attenuation was also investigated with lead
plates. The theoretical equation M = M0 e−µt was found to fit the
experimental data quite well, but a different equation, M  M 0 e – kt ,
was found to provide a much improved fit, especially where the
standard thickness lead plates were not combined to provide addi-
tional thicknesses to test.

Introduction

History of radioactivity
Antoine Henri Becquerel (1852 - 1908)
In March of 1896, a French physicist, Antoine Henri Becquerel,
placed some wrapped photographic plates into a draw along with
some crystals containing uranium. Much to his surprise, when he
returned to the draw sometime later he discovered the plates were
exposed. He concluded that the uranium crystals must have been
emitting something, since the exposure had occurred without the
presence of the usual initiating energy source - sunlight. Becquerel,
though, did not pursue his discovery of radioactivity any further.
This was left to the Curies.
226 • Physics Lab Experiments

Marie (1867 - 1934) and Pierre (1859 - 1906) Curie

Becquerel had already discovered that the “ray” emitted by ura-
nium could turn air into a electrical conductor. The Curies followed
this up, testing various materials. On February 17th 1898 they discov-
ered that an ore of uranium, pitchblende, produced a current over 300
times more than that produced by pure uranium. After repeating the
experiment and obtaining the same answer, they concluded that some
very active substance other than uranium must exist within the pitch-
blende. In a paper detailing their experiment, they hypothesized the
existence of a new element that they named polonium, and coined the
phrase “radioactive” to describe elements like uranium and polonium.
Continuing their work the Curies discovered another radioactive ele-
ment, radium. Marie continued her work after Pierre’s death and dis-
covered there was a decrease in the rate of radioactive emissions over
time and that this decrease could be calculated and predicted.
In 1903, Becquerel and the Curies together received the Nobel
Prize in physics. This award was for their discovery of radioactivity
and their other contributions in this area.
Ernest Rutherford (1871 - 1937)
Up until Ernest Rutherford discovered the structure of the atom
in 1911, there was very little scientific explanation for what caused
radioactivity and how the emissions affected the original atoms.
In a series of experiments he bombarded a piece of gold foil with
alpha particles emitted by a radioactive material. Most of the par-
ticles passed through the foil undisturbed, suggesting that the foil
was made up mostly of empty space rather than of a sheet of solid
atoms. Some alpha particles, however, bounced back, indicating the
presence of solid matter. Rutherford concluded that atoms consisted
primarily of empty space surrounding a well-defined nucleus.
Many more years passed until the full effect of radioactive decay,
on the atoms undergoing the decay, was fully understood.
Quantitatively measuring the rate of radioactive decay
The radioactivity of a substance is measured by how many decays
occur in a unit time. Both Becquerel and the Curies have units of radio-
activity named after them. The Curie (Ci) is defined as 1Ci = 3.7 ×1010
decays sec−1 and the Becquerel (Bq) is defined as 1 Bq = 1 decay sec−1.
 Common Experiments • 227

Modern Day Uses for Radioactive Isotopes

Today, radioactive isotopes are used widely in industry for thick-
ness control and other forms of testing; in health care as tracers,
cancer treatments and in the sterilisation of surgical instruments; in
domestic appliances such as smoke detectors; and for power produc-
tion on spacecrafts, navy vessels, and power stations.
Aims of this experiment
This experiment is designed to test the background statistical
theory behind radioactive decay and investigate one set of appa-
ratus used for measuring radioactivity - the Geiger Muller tube.
It also seeks to investigate some of the properties of radioactivity
such as exponential decay and the ability of certain materials to
scatter/absorb particular types of radiation. These and other prop-
erties are made use of in many of the processes mentioned in Mod-
ern Day Uses for Radioactive Isotopes.

Theory

Binomial Distribution
Radioactive decay is a random process, so the probability of any
particular atom decaying is independent of whether any other par-
ticular atom has decayed. This is characteristic of the Binomial dis-
tribution.

P  r   n Cr pr q n–r

If an atom has probability t of decaying in a short time interval t,

given it has not decayed already, the probability distribution func-
tion for survival is (for independent probabilities)

F  t  dt   F  t  . 1 – ldt  : (6.22)
228 • Physics Lab Experiments

Where F(t+t) is the probability of survival at time t + t , F(t) is

the probability of survival at time t and (1 − t) is the probability it
doesn’t decay in next t. If a Taylor expansion is performed on the
left hand side of Equation (6.22) it becomes:

dF
F t  d t  ...  F  t  . 1 – ld t  : (6.23)
dt

As t  0, (6.23) becomes:

dF
 – lF  t  : (6.24)
df

Equation (6.24) is a 1st order differential equation for F(t), which

has the solution:

F  t   Ce – lt : (6.25)

Since F(0)=1 then C=1 i.e., the particle had not decayed at t=0.
Equation (6.25) can be multiplied by N0, the total number of atoms
present to give:

N  N0 e – lt
: (6.26)

This tells us the number of particles, N, left after time t.

Normal and Poisson Approximations to the Binomial
Approximations to the Binomial can be evaluated using the Nor-
mal or the Poisson distribution. The Normal approximation (when n
and np are large) is:

1
e 
– r – np  2 npq 
2 2
P  r   n Cr pr q n–r 
2pnpq

The Poisson approximation (when n large and p small) is:

 Common Experiments • 229

 np 
r
P  r   n Cr pr q n–r  e –np
r!

Chi Squared Test

In order to see how well a theoretical distribution fits a series of
experimental results a Chi Squared test can be performed. A good
fit requires that 2  v where v is the number of degrees of freedom.

 ytheoretical – yexperimental 
2

x 
2
 ytheoretical

The reduced value of 2 which can be looked up in tables to give a

probability is

x2
x v2 
v

If this probability is greater than 0.05 the experimental data can be

said to fit the theory.
The Geiger Muller Tube
In order to measure the decay rate of the radioactive sources a
Geiger Muller Tube was used: see Figure 2.2.

Figure 6.10: The Geiger Muller Tube.

230 • Physics Lab Experiments

A large potential difference is applied by the EHT power sup-

ply, so as the tube becomes negatively charged with respect to the
wire. This (electric) potential difference is decided with reference
to a number of conditions. It needs to be above the threshold fre-
quency, below the (electric) potential difference that causes electrical
discharge through the gas and on the plateau of the calibration graph.
When any form of radiation enters the tube thorough the thin mica
window it ionizes a gas atom. The electron created moves towards the
central wire and the gas ion moves towards the tube (due to electro-
static attraction). The ion moves more slowly than the electron since it
is move massive i.e. by Newton’s second law its acceleration is lower.
Due to the high EMF of the EHT supply, the electron is accelerated
rapidly and causes the ionization of another atom. This builds rapidly
into an avalanche of electrons. The velocity of propagation of the elec-
trons is high enough so that as a result of the avalanche of electrons
the charge available for collection by the wire has a constant value,
irrespective of where in the tube the original ionization occurred.
When the electrons from the avalanche have been collected the
slowly moving sheath of positive ions acts as a partial electrostatic
shield and therefore reduces the field below that necessary for ion-
ization by collision. This avalanche of electrons has the effect of pro-
ducing a large current in the external resistance. This can then be
used to increment a counter for each particle/ray reaching the tube.
The tube is now ready to make another count.
The sheath of positive ions can eject electrons from the tube,
leading to more avalanches and multiple counts for only one decay.
Alcohol is usually added to the gas in the tube to quench this extra
discharge. The alcohol does this since it has a lower ionization energy
than the gas, and so is ionized first. The energy that was used to release
extra electrons is now used in dissociating the alcohol molecule.
Dead Time
Geiger Muller tubes have limitations. One important limitation
in this experiment is dead time. This is caused as an after effect of
the electrical breakdown of the gas filling the Geiger Muller tube.
Since the tube will not register any other counts for a short period
of time after each count, a significant proportion of the overall count
can be lost.
 Common Experiments • 231

If two identical sources are used and N1 is the number of counts

with only source 1 present, N2 is the number of counts with only
source 2 present and N3 is the number of counts with both source 1
and 2 present then the dead time can be shown to be:

1 1 N 2  N1 – N 3
t – –
N3 N 32 N 1N 2N 3

Proof of this equation is given in Appendix.

Isotopes and Types of Decay
Atoms consist of protons, neutrons, and electrons. In order to
have a zero charge an atom must have the same number of protons
and electrons. Atoms of different elements have different numbers
of protons. Some elements have more than one isotope - i.e., they
have a different number of neutrons (but the same number of pro-
tons). Some isotopes are radioactive and decay, in order to become
more stable, in one of the following ways.
Alpha Decay - The nucleus loses two neutrons and two protons.
Beta Minus Decay - A neutron (from the nucleus) decays to
become a proton; and an electron and an anti-electronneutrino that
are both emitted.
Beta Plus Decay - A proton (from the nucleus) decays to become
a neutron; and a positron and an electron neutrino that are both
emitted.
Gamma Rays - Can be created in a number of ways as the atom
decays, but the method relevant to the sources used in this experi-
ment is the annihilation of an electron from the air and a positron
emitted during beta plus decay.
Shape of Calibration Graph
The counting starts at a threshold (electric) potential differ-
ence indicated by A on Figure 6.11. This is when the field becomes
strong enough so that ionization of the gas atoms occurs. The count
rate then rises rapidly up to the plateau beginning at B. The Geiger
Muller tube is operated on the plateau at the (electric) potential dif-
ference B, since any slight variation in the (electric) potential differ-
ence produced by the (electric) potential difference supply will then
have a negligible effect on the recorded counting rate. The count
232 • Physics Lab Experiments

Figure 6.11: Idealalized Calibration Graph

rate rises rapidly at higher (electric) potential differences since

breakdown has occurred i.e., the quenching of the alcohol is not
enough to stop multiple counts for a single decay. Ideally the plateau
should be flat, but in practice it slopes with a small positive gradient
since as the (electric) potential difference increases the probability
of spurious counts also increases.

Experimental Methods

Sources Used
All these sources have very long half lives so as the count rate
will be effectively constant over the whole period of the experiment.
 Common Experiments • 233

Finding the Optimal Voltage at which to run the

Geiger Muller Tube
In order to determine the (electric) potential difference, B, at
which to run this Geiger Muller tube, measurements were taken
of the count rate of the 90Sr source and EHT EMF. The computer
was set to take the number of counts over a 10 second period, at
(electric) potential differences from 0 V to 1200 V in 10 V steps. The

Diagram of Apparatus

Figure 6.12: The apparatus used

234 • Physics Lab Experiments

count rate was calculated by dividing the count by the number of

seconds. The Calibration Graph section shows the graph obtained
by plotting the (electric) potential difference against count rate.
Shape of the Normal Approximation
From the Normal and Poisson Approximations to the Binomial
section, it is noticed that for a Normal approximation to a Binomial
distribution to be accurate, n and np must be large. This can be
achieved by measuring a large number, n, of count rates. Since np
is the mean of the Normal distribution, the mean count rate should
also be large - around 100 should be sufficient. The mean count rate
can be varied by adjusting the position of the source in the chamber.
This time the computer was set to take 1000 readings of the
count over 1 second (which gives the count rate). The mean and
variance of these results was calculated using the standard for-
mulae – see Lyons (1991). The theoretical Normal distribution
with this mean and variance was calculated and then multiplied
by 1000 so as the theoretical and experimental graphs could be
plotted and compared on the same scale. The section Graph of the
Normal Approximation shows the graph obtained.
Shape of the Poisson Approximation
Again from the Normal and Poisson Approximations to the
Binomial Section, it is noticed that for a Poisson Approximation
to a Binomial distribution to be accurate, n must be large and p
must be small. In order to ensure p was small, a scatter/absorber
material was inserted between the source and the tube so as the
mean was reduced to around 2 – 4 counts/second. Again, the com-
puter was set to take 1000 readings of the count rate over 1 second.
The mean of these readings was calculated. The theoretical Pois-
son distribution with this mean was then calculated and multiplied
by 1000 so as the theoretical and experimental graphs could be
plotted and compared on the same scale. The Graph of the Poisson
Approximation section shows the graph obtained.
Chi Squared Test
For the Normal and Poisson approximation the values of 2 and
prob(2) were calculated. This enables the closeness of the fit to be
analyzed quantitatively.
 Common Experiments • 235

Background Count
The background count (i.e., the count rate with no source
present) was measured and subtracted from all subsequent mea-
surements. The count rate was found to be 0.48727 counts sec−1.
Dead Time Measurement
In order to calculate the dead time, t, values of N1, N2 and N3
were obtained. A value for N1 was obtained by performing a series of
measurements of the count over a period of 10 s with just one 204Tl
source present. A value for N2 was obtained by the same method with
the other 204Tl source present. Finally, a value for N3 was obtained by
the same method with both the sources present. Putting these values
into the equation found in the earlier section called Dead Time gives
a value for the dead time. Later, we will see the results and final
value for dead time.
Dead Time and the Background Count Rate
The background count was adjusted for the dead time. The
adjusted background count rate was now found to be 0.48739 counts
sec−1. This new value was subtracted from all subsequent results.
The Inverse Square Law
Since the particles emitted by a radioactive source are random
in both time and direction, the count rate should fall with the square
of the distance between the source and the Geiger Muller tube.
The experiment was performed by placing the source on each of the
shelves in the chamber and then taking a reading of the count rate over
a 10 second period. The two sources tested were 90Sr and 22Na. The
intershelf distance was constant and found to be 12.8 mm. Graphs of
1 M , where M is the count, against distance could then be plotted.
Using the intershelf distance method meant that there was no
need to know the absolute distance between the source and the
Geiger Muller tube window. In order to plot the results the distance
of the first shelf was arbitrarily chosen to be 1m. This method still
allowed for comparison of the results and for the inverse square law
to be tested.
After the points were plotted and a line of best fit was added,
Excel was used to give the equation of the line and also a value of
236 • Physics Lab Experiments

R2. From the equation of the line the x-axis intercept was also calcu-
lated. The results are given in the Inverse Square Law section.
Exponential Attenuation
The absorption of particles by a certain material can be inves-
tigated by placing the source at a constant distance from the Gei-
ger Muller tube and inserting various thicknesses of the material
in between. Alpha and Beta particles lose energy progressively
and have a definite range in a certain material but gamma rays are
expected to be absorbed exponentially.
In order to test whether this was the case, the 22Na source and
various thicknesses of lead were used. These plates were inserted in
various combinations in between the source and the Geiger Muller
tube and the count rate measured. The results are provided in the
Appendix.

From preliminary graphs it was concluded that the gamma were

absorbed exponentially. Kaplan (1956) gives the equation M=M0e−µt
where M is the count rate, M0 is the count rate with no lead being
used, t is the thickness of the material and µ is the linear attenuation
coefficient. By comparing

In M = In M 0  mt

y  c  mx
 Common Experiments • 237

it can be seen that if a graph of lnM against t is plotted with the inter-
cept fixed at lnM0, the gradient of this graph would be the constant
−µ . Graphs are shown in the Exponential Attenuation section.

Observations and Results

Calibration Graph
The threshold (electric) potential difference (A) can be seen to
be 650 V and the start of the plateau (B) is at 750 V. For the rest of
the experiment the Geiger Muller Tube was run with a EMF supply
of 750 V.

Figure 6.13: Calibration Graph

238 • Physics Lab Experiments

Graph of the Normal Approximation

Figure 6.14: The Normal Approximation

Figure 6.15: The Poisson Approximation

 Common Experiments • 239

The bars show the experimental results and the line shows the
theoretical results. The mean is 109, the variance is 10.2, the prob(2)
is 0.47 and the 2 value is 73.1.
Graph of the Poisson Approximation
Again the bars show the experimental results and the line shows
the theoretical results. In order to sufficiently reduce the mean
count rate a piece of aluminium 0.032 inches thick was found to be
necessary. The mean is 3.5, the variance is 3.5, the prob(2) is 0.61
and the 2 value is 7.29.
The Nx values shown above are already corrected for the back-
ground count and then the dead time was calculated. The mean,
standard deviation and standard error on the mean were calculated
240 • Physics Lab Experiments

Figure 6.16: The Beta Source

- see Lyons (1991). The mean is 510 µs and the standard error on
the mean is 70 µs.

Inverse Square Law

Beta Source
With the beta source, the x-axis intercept was found to be at
-0.19 and the R2 value was found to be 0.9977.
 Common Experiments • 241

Gamma Source

Figure 6.17: The Gamma Source

Exponential Attenuation

Figure 6.18
242 • Physics Lab Experiments

Figure 6.19

With the gamma source, the x-axis intercept was found to be at -0.52
and the R2 value was found to be 0.998.

Figure 6.20
 Common Experiments • 243

Figure 6.21

Figure 6.22

Using Graph 4.6.1 µ was calculated to be 1.1974. Using:

m s

r Ma
244 • Physics Lab Experiments

from Kaplan (1956), where µ is the linear attenuation coefficient, 

is the average density of the lead,  is the cross sectional area of a
scattering center (a lead atom) and Ma is the mass of a lead atom.
 was found to be 3.33×10−23 cm2. Since 1 barn =1×10−24 cm2,  =
33.32 barns.

Discussion

The plateau on the calibration graph was found to start at around

750 V. All readings for the remainder of the experiment were carried
out with the Geiger Muller tube running at this (electric) potential
difference.
The count rate was found to fit the Poisson distribution more
closely than the Normal. The prob(2) value for the Poisson was 0.61
where as for the Normal it was only 0.47. Considering the conditions
for both these approximations outlined in the Normal and Poisson
Approximations to the Binomial section, the number of count rates
measured should have been higher than 1000. This should ensure
the conditions are satisfied more exactly and should increase the
closeness of fit between the theoretical and experimental results.
The dead time was found to be 490±70 µs. This value is of the
order of that expected for a typical Geiger Muller tube of the type
used. Therefore this can be considered to be a reliable value. Per-
haps, in order to increase the accuracy of this part of the experi-
ment more readings could be taken, so as to decrease the error, and
increase the certainty of the measurement.
Graphs 4.5.1 and 4.5.2 for the beta and gamma emitting sources
respectively, very closely obey the inverse square law. This is shown
by the closeness of the R2 value to 1. The x axis intercept for the
gamma source is lower than that for the beta source, suggesting
that the gamma rays, as expected, are penetrating further into the
Geiger Muller tube than the beta particles. This could indicate
that the Geiger Muller tube was acting as a sensitive volume in the
case of the gamma source and a sensitive area in the case of a beta
source (beta particles being much less penetrative). There seemed
to be little trouble with the beta particles not having enough energy
to penetrate the mica window of the Geiger Muller tube since a
 Common Experiments • 245

significant count rate was always obtained at distances comparable

with the gamma source.
Kaplan (1956) gave the equation for exponential attenuation as
M=M0e−µt where M is the count rate, M0 is the count rate with no
lead being used, t is the thickness of the lead and µ is the linear
attenuation coefficient. Looking at Graph 4.6.1, where the intercept
was fixed at lnM0 the data fits the equation very well. It provides an
even better fit if the intercept is not fixed, as shown in Figure 6.18.
While taking the first few sets of data and plotting some preliminary
graphs it did not look like the data was going to fit the equation from
Kaplan (1956). Figure 6.21 shows the data for the first 4 combina-
tions of the lead tested. (These were all single pieces of lead of the
specified thickness. In all further trials, a combination of lead plates
was used and the total thicknesses was simply taken to be the sum of
the individual thicknesses.) This inaccuracy led a search for another
equation that the data fitted more accurately. Using the relationship
M  M 0e k t and plotting (lnM0 − lnM)2 against t, with the same four
results as before, there appeared to be an almost perfect straight line
fit as Figure 6.20 shows. Using the modified equation produced an
R2 value of 0.9987, this is much closer to 1 than the R2 for the graph
using the equation from Kaplan (1956) which was 0.729.
After performing further tests with other thicknesses of lead the
whole set of results was plotted using each of the above equations.
Figure 6.22 shows the whole set of data plotted with the modified
equation and Figure 6.18 shows the whole set of data plotted with
the original equation from Kaplan (1956). The modified equation
still produces a better fit with a R2 value of 0.9475 compared to
0.8424. (The graph plotted for the modified equation, should go
through the origin, so the line of best fit was fixed to do this. This
graph for the equation from Kaplan (1956) should go through ln M0
so it was fixed to do this. Therefore a valid comparison can be drawn
between these two graphs and not between Figure 6.19 where the
intercept was not fixed. However fixing intercepts potentially dis-
guises any systematic error which is present in the results.)
Further research would be needed to determine why the modi-
fied equation is especially appropriate when single blocks of lead are
used. It would be interesting to test with other thicknesses made up
solely of one block of lead to see if the modified equation is indeed
246 • Physics Lab Experiments

better or if the closeness of fit of the modified equation is an experi-

mental one off. As the number of results increased the fit of the
Kaplan (1956) equation improved and the fit of the modified equa-
tion deteriorates slightly, but in the limit of the results taken, still
remains more accurate.

Conclusions

The start of the plateau for the Geiger Muller tube was found to
be 750 V. Both the Normal and Poisson distribution are valid approx-
imations to the Binomial in this case, although the Poisson provides
the closer fit between experimental data and theory. The dead time
of the Geiger Muller tube was calculated to be 490±70 µs. The count
rate for both the gamma and beta sources was found to accurately
obey the inverse square law. The data gathered while investigating
exponential attenuation supports the equation, M=M0 e−µt given in
Kaplan (1956), but the modified equation, M  M 0e k t fits the data
more accurately. The cross section of a scattering center was found
to be 33.32 barns.

References

Taylor, Denis (1957) The Measurement of Radio Isotopes,

Methuen + Co.
Burcham (1963) Nuclear Physics: An Introduction, McGrawhill
Book Company
Kaplan (1956) Nuclear Physics, Addison-Westley
Halliday, Resnick, Krane (1992) Physics, Wiley (Chichester)
Boas, Mary (1983) Mathematical Methods in the Physical Sci-
ences, Wiley (Chichester)
Lyons, Louis (1991) A Practical Guide to Data Analysis for
Physical Science Students, Cambridge University Press (Cambridge)
 Common Experiments • 247

Barford, W. C. (1985) Experimental Measurements: Precision,

Error and Truth, 2nd Edition, Wiley (Chichester)

Appendix

Dead Time
If N counts are recorded in 1 second, and the tube is dead for a
time t after each count then in 1 second the tube will be dead for Nt
seconds. If an ideal counter has M counts in 1 seconds then

N  M  MNt

N  M (1  Nt )

If M1 is the count rate with only source 1 present, M2 is the count

rate with only source 2 present and M3 is the count rate with both
sources 1 and 2 present then:

M 3  M1  M 2

N3 N1 N2
 
 1 – N 2t   1 – N 1t   1 – N 2t 
Simple rearranging yields a quadratic in t:

0 = N 1 N 2 N 3 t 2  2 N 1 N 2t  (N 2  N 1  N 3)

The quadratic formula gives two results. A simple approximation to

the quadratic in t can be made by assuming that since t is small t2 
0. Thus giving the equation:
248 • Physics Lab Experiments

N 2  N 1  N 3  2 N 1 N 2t

The solution to these two equations in can be compared, and it can

be determined that the minus version of the quadratic formula gives
the correct dead time.
Exponential Attenuation Results
Common Experiments • 249
250 • Physics Lab Experiments
 Common Experiments • 251

6.2.3 Example Report: Latent Heat of Vaporization of

Liquid Nitrogen

Abstract
The latent heat of vaporization of liquid nitrogen was calcu-
lated by measuring the boil-off from an insulated dewar flask when
a known amount of heat was supplied. It was found to be signifi-
cantly higher than the accepted value of 199 kJkg−1 at 249 kJkg−1.
The high value can be accounted for because the volume measuring
was stopped as soon as the heater was switched off, and not contin-
ued as the heater cooled. The specific heat capacity of a range of
materials was then calculated over two temperature ranges, 293 K
252 • Physics Lab Experiments

to 77 K and 87 K to 77 K. The results, though high in comparison

with the accepted values were explainable by the high latent heat of
vaporization obtained in the first part of the experiment, are quali-
tatively consistent with Debye theory i.e., as the temperature drops
the specific heat capacity drops too.

Introduction

Latent Heat of Vaporization

The latent heat of vaporization is the energy required to change
the state of a material from a liquid to a gas at constant temperature.
Specific Heat Capacity
The specific heat capacity of a material is the energy required
to raise the temperature of one mole of the material by one Kelvin.

History of Latent Heat

Joseph Black (1728 - 1799)

While at the University of Glasgow, Black was performing
research on the freezing and melting of water and water/alcohol
mixtures that led to the concept of latent heat of fusion. Similar work
he performed later led him to establish the concept of latent heat of
vaporization - eventually leading to the general concept of specific
heat capacity (c.1760).
Sir James Dewar (1842 - 1923)
While working on cryogenics at the Royal Institute between
1877 and 1904 Dewar invented the Dewar flask. The basic con-
struction of the Dewar flask consists of a double walled glass/metal
container with the gap between the walls “filled” with a vacuum. In
1878, he achieved the liquefication of oxygen (boiling point 97 K). In
the closing years of the 19th century he liquefied hydrogen (boiling
point 20 K). His pioneering work on low temperature physics and
vacuum techniques paved the way for a wide range of experimental
 Common Experiments • 253

work at very low temperatures. From 1904 to 1914, he extended his

work to include low temperature calorimetry investigations.

History of Specific Heat Capacity

Dulong and Petit

Dulong and Petit developed the first theory behind specific heat
capacity in 1819. This treated the atoms in a crystal as classical oscil-
lators with kT/2 of energy per degree of freedom. This theory pre-
dicted the specific heat capacity to be 3R per mole. Although this
is quite accurate at room temperature for many materials, it fails
badly at low temperatures and even fails for some materials (e.g.,
diamond) at room temperature.
Einstein
Einstein observed that the specific heat capacity of a material
changes with temperature and that not all materials had the same
specific heat capacity at room temperature as Dulong and Petit had
predicted. To explain this he developed a theory in which he viewed
the atoms in a crystal as independent quantum harmonic oscillators,
oscillating at one fixed frequency. This approach, though much more
accurate than the Dulong and Petit method still failed at very low
temperatures. Experimental results showed the dependence of spe-
cific heat capacity on temperature was T3 rather than exponential as
predicted by the Einstein approach.
Debye
The modern day theory was developed by Debye, who extended
Einstein’s approach by assuming that the oscillators were coupled
and could vibrate at a whole range of different frequencies.
The Importance of Latent Heat and Specific Heat Capacity
The knowledge of the latent heat of vaporization of material
allows many calculations to be performed. The Clausius-Clapeyron
equation is just one such equation that makes use of latent heat.
Latent heat can also be used to calculate how much energy is needed
to raise the temperature of a material from T1 to T2 when a change
254 • Physics Lab Experiments

of state occurs between the two temperatures. The development of

a theory of specific heat capacity has given a detailed understanding
of how atoms in a crystal behave as temperature changes and led to
the most appropriate material being used in many cases where the
thermal properties of a material are an important consideration.
Aims of this Experiment
This experiment is designed to calculate the latent heat of vapor-
ization of liquid nitrogen and the specific heat capacity of a range of
samples over two different temperature ranges. First, from 293 K
(room temperature) to 77 K (boiling point of liquid nitrogen) and
secondly over the range 87 K (boiling point of liquid argon) to 77
K. It also aims to qualitatively test the Debye theory that says that
specific heat capacity decreases as temperature decreases.

Theory

Latent Heat of Vaporization

The latent heat of vaporization of a material is the energy
required to change the phase of the material from a liquid to a
gas at a constant temperature. In order for the phase change to
occur energy must be supplied to the system. The energy taken
in during the phase change does not go to increase the tempera-
ture of the material. Instead, this energy is used to provide the
increased potential energy of the gas phase (i.e., overcoming
attractive forces between the molecules) and, when the phase
change results in expansion, to do external work in pushing back
the atmosphere.

Q
Lvap 
m

where Lvap is the latent heat of vaporization, Q is the energy input

and m is the mass of the sample that changes phase. The units used
for are kJkg−1.
 Common Experiments • 255

Heat Capacity
The heat capacity, CT, is defined as the ratio of the energy input
by heating and the resulting temperature change (where Q is the
energy input and T is the temperature change):

Q
CT 
T

In the limit of small Q and T, this becomes:

dQ
CT 
dT

This is an extensive parameter, dependent on the size of the sample.

The units used for CT are JK−1.
Specific Heat Capacity
The specific heat capacity of a material is an intensive param-
eter, not dependent on the size of the system:

CT
C
n

where n is the number of moles of material. The units used for are
JK−1mol−1.
Debye Theory
The classical theory explaining the specific heat capacity of a
solid assumes that in a crystal lattice there can only be vibrational
energy - there is no rotation or translational energy. Each atom on
a lattice site acts as an oscillator in three independent directions, so
there are 3N oscillators (where N is the number of atoms in the lat-
tice) each with two degrees of freedom (potential energy and kinetic
energy). Classically there is kT/2 of energy per degree of freedom
(where k is the Boltzman constant and T is the temperature) so:

1
 3N  2  2 kT  3NkT
256 • Physics Lab Experiments

Figure 6.23: Graph to show the variation in CV with temperature, T.

From the Specifc Heat Capacity section

dQ
CV   3R
dT

where CV is the specific heat capacity per mole at constant vol-

ume and R is the gas constant. This gives the Dulong and Petit
value for the specific heat capacity. However, CV is dependent
on temperature, and clearly not independent as Figure 6.23 sug-
gests:
In an improved approach Einstein viewed the lattice as an array
of independent quantum harmonic oscillators with fundamental fre-
quency f. The Einstein temperature, where h is Plank’s constant and
E is the Einstein temperature, is defined as:

hf
qE 
k

and
2
q  e qE T
CV  3Nk  E 
T 
 eq 
T 2
E
–1
 Common Experiments • 257

At high temperatures this gives the classical result, but at low

temperatures it gives:
2
q 
CV  3Nk  E  e –qE T
T 

The Einstein approach gives a much better fit to experimental data

than the classical (Dulong and Petit) approach, but there is still sig-
nificant error, especially at low temperatures where CV  T3 and for
metals at high temperatures. Einstein assumed that the oscillators
are independent of each other; clearly this can’t really be the case if
the atoms are locked in a lattice structure. He also failed to take into
account any directionality in the lattice structure and assumed there
was only one possible vibrational frequency.
Debye recognized that the oscillators are not independent, but
coupled, so that there are a range of vibrational frequencies cor-
responding to standing waves in the crystal structure. The distur-
bances can travel around the material in any direction.
The density of available states (or frequencies) is:

12pV f 3
D D ebye 
c s3

where V is volume and cs is the speed of sound. Since the most vibra-
tions which can occur is 3N, the maximum number of states is also
3N:

f max 12pV f 2

0 c s3
df  3N

so:

3 3N c s3
f max
4pV

The maximum frequency defines a characteristic temperature, the

Debye temperature, D where a=(V/N)1/3 is the average interatomic
spacing:
258 • Physics Lab Experiments

k qD  hf max

13
 3 hc s
qD   
 4p  ka

Quantum theory restricts the energy of a sound wave to increments

of hf. With each extra hf of energy it is said that an extra phonon
(the quantum of elastic energy in a lattice) is present. The estimated
number of phonons in the mode with frequency f is:

1
n f   hf kT
e –1

The energy present in the lattice is then:

f max
E   0
hf D D ebye n  f  df

f max 12phV f3
E  0 c s3 e hf kT – 1
df

At low temperatures, T  D, this becomes:

4
12phV  kT  qD T x3
E    
c s3  h  
0
x
e –1
dx

where x=hf/kT. So:

 E 
3
12p 4 T 
CV     Nk  
 T  V 5  qD 

At high temperatures T >> D it becomes 3NkT and CV=3R per

mole.
CP (specific heat capacity at constant pressure) is the most appro-
priate quantity to use in this experiment since atmospheric pressure
will remain constant. But since the solids tested here contract only
 Common Experiments • 259

very little compared to gases there is no need to draw a distinction

between CP and CV.
Energy Dissipated by a Resistor
By definition, the potential difference, V, between two points is
equal to the work done, W, in moving a positive charge, Q, from the
lower potential to the higher potential. Thus:

W  QV

If the current, I, is constant (where t is time):

Q  It

So:

W  It (V2 V1)

If V1=0, then the energy dissipated by a resistor (i.e., a heater) is:

W  VIt

Number of Moles of a Gas

The gas volume meter records the number of liters of gas pro-
duced. In order to perform calculations using this result the volume
is needed in terms of numbers of moles, not number of liters.
Using the equation of state of an ideal gas:
260 • Physics Lab Experiments

PV  nRT

where P is pressure, V is volume, n is number of moles, R = 8.31

JK−1mol−1 is the gas constant and T is the temperature, the volume
of one mole of gas at room temperature (20°C) and pressure (101325
Pa) is found to be 24 dm3. (This theory assumes that the nitrogen gas
has reached room temperature by the time it has traveled through a
long tube and goes into the gas volume meter.)
Error Calculations
Errors were calculated on all results using the formulae given in
Lyons [7] for the propagation of errors. If:

a b  c

then

sa2  sb2  sc2

and if: v

x
f  xy or f 
y

then:
2 2 2
 sf   sx   s y 
 f    x    y 
 

where k is the standard deviation of the variable k.

 Common Experiments • 261

Experimental Methods

Latent Heat of Vaporization of Liquid Nitrogen

Figure 6.24: The Latent Heat Calorimeter. A diagram of the apparatus used in the Latent
Heat of Vaporization Liquid Nitrogen section.

In order to calculate the latent heat of vaporization of liquid

nitrogen a large dewar flask was filled with liquid nitrogen. The cal-
orimeter was carefully inserted into the large dewar flask and then
also filled with liquid nitrogen. The heating coil was then inserted
into the calorimeter. The apparatus was allowed to cool down. Dur-
ing the initial cooling of the apparatus both the large dewar flask and
262 • Physics Lab Experiments

the calorimeter were topped up with liquid nitrogen as the boil off
was considerable. When the boil off had reduced to a steady and low
rate, the experiment was started.
Firstly a background rate was taken, so as this contribution could
be subtracted from the results so that the contribution from the
heating coil alone could be calculated.
The heating coil was then turned on, and the current and (elec-
tric) potential difference were recorded using a multimeter. As the
liquid nitrogen in the calorimeter boiled off the volume was recorded
from the gas volume meter.
Using the current and (electric) potential difference supplied to
the heater and the time period over which the measurements were
taken, the heat dissipated by the heater was calculated. The volume
of N2 evolved as a result of this amount of heat was read from the gas
volume meter. This was converted into the number of moles as by
dividing by 24 (as outlined in the Number of Moles of a Gas section).
The mass of N2 produced was calculated using:

m
n
Mr

where n is the number of moles, m is the mass in grams and Mr is the

atomic mass in grams (28 in the case of N2). Lvap was then calculated.

Specific Heat Capacity Over Temperature Range 293 K

 77 K
The heating coil was replaced by a stainless steel tube with a
bladder and rubber hose attached. The bladder is needed to help
prevent air condensing inside the tube. Air is mainly made up of
nitrogen and oxygen. In the stainless steel tube submerged in liquid
nitrogen, oxygen can condense since its boiling point is 97 K and that
of nitrogen is 77 K. As well as producing undesired inaccuracies in
the experiment, having liquid oxygen present is a fire hazard.
The most effective way to prevent oxygen condensing is for
it to not be present at all in the stainless steel tube. This can be
achieved by creating an atmosphere of nitrogen inside the tube.
Firstly the tube was partially cooled in the calorimeter so that
 Common Experiments • 263

the next part could be performed more easily. The tube was then
removed from the calorimeter and a little liquid nitrogen was
added to it. The end of the rubber hose was corked so as to seal
the tube. As the liquid nitrogen boiled off, the bladder enlarged.
While the tube and hose were held upside down with the cork
removed the bladder was squeezed and the nitrogen gas expelled.
This was repeated until all the nitrogen had boiled off (i.e., the
bladder stopped enlarging when the hose was re-corked). Since
oxygen gas (1.43 kgm−3) is denser than nitrogen gas (1.25 kgm−3),
is sinks out of the tube, encouraged by the nitrogen, thus leav-
ing an oxygen-free atmosphere inside the stainless steel tube. The
hose was then immediately re-corked. This method proved very
successful at preventing the formation of liquid oxygen in the
stainless steel tube.
The stainless steel tube was then reinserted into the calorimeter
and any significant boil off from the liquid nitrogen in the calorim-
eter or the large dewar was topped up. When the boil off had settled
to a low steady value a background reading was started.
After the background reading was taken the cork was briefly
removed and the sample was added to the hose while squeezing the
bladder so as no air could enter the tube or hose. The hose was then
immediately re-corked. The sample dropped into the bottom of the
stainless steel tube and readings were taken from the gas volume
meter every 30 seconds until the rate of boil off had dropped below
that of the background rate.
The sample was not added to the hose earlier and left there
while the background reading was taken (with the hose bent over
to prevent the sample entering the calorimeter) since the ambient
air temperature inside the hose was not the same as the ambient air
temperature in the room. This would have meant the temperature
range over which the measurements were taken would not have been
known.
Each time a reading was taken a new background reading was
taken since it was more accurate than trying to keep the level of the
liquid nitrogen in the calorimeter and the dewar constant between
readings.
This part of the experiment was performed with samples of
copper, aluminium, graphite, zinc, lead, and rock salt.
264 • Physics Lab Experiments

Specific Heat Capacity Over Temperature Range 87 K 

77 K
The same procedure was followed as in the Specific Heat Capac-
ity Over Temperature Range 293 K → 77 K section but this time the
samples were first cooled in liquid argon (boiling point 87 K) before
being inserted into the stainless steel tube. The liquid argon was made
by passing argon gas through a copper tube submerged in liquid nitro-
gen. The samples were attached to a piece of thread so as they could
be removed from the liquid argon quickly and easily and inserted in to
the stainless steel tube. In the light of the previous work an improved
method for taking a background reading was introduced. This con-
sisted of reading the gas volume meter every 30 seconds for 600 sec-
onds and then inserting the cooled sample and taking another set
of readings every 30 seconds for 600 seconds. This provided a more
accurate representation of the background reading than earlier where
it was assumed to be constant. Clearly, the calorimeter can’t be per-
fectly insulated, especially with a large conducting metal tube sticking
out into the air, so the background reading can’t be exactly constant.
Volume and Density
All the samples were weighed using an accurate balance. The
volume of the samples was calculated using a measuring cylinder
half full of distilled water. The volume was recorded before and after
the samples were placed in the water. The volume of the sample is
then the difference in these volumes. The density was then calcu-
lated using the mass and volume of the sample.
Safety
While performing this experiment careful consideration was
given to safety. Both the liquid nitrogen and liquid argon used are
obviously very cold. Goggles were used to prevent splashes entering
the eyes. Gloves were also worn when handling the apparatus as it
was very cold. The liquids were added very slowly and carefully to the
apparatus (which was at room temperature initially) to prevent vio-
lent boil off. It was also noted that the extreme cold causes glassware
to become brittle, so glass-glass and glass-metal contact was avoided
as much as possible. The spacer holding the calorimeter in the large
dewar has a loose fit to allow any excess gas to escape so as pressure
 Common Experiments • 265

does not build up. The possible formation of liquid oxygen (highly
flammable) was avoided by using the method outlined in the Specific
Heat Capacity Over Temperature Range 293 K → 77 K section.
Calculation of Results Over Temperature Range 293 K
 77 K
The volume of N2 evolved over the previous 30 seconds was
calculated from the results. This was then plotted against time to
produce a graph showing the rate of N2 boil-off (see Table (6.1)
for an example of this graph). The background reading was then
added to the same graph. At the point where the two lines meet, it
is assumed that the sample is at 77 K (the same temperature as the
liquid nitrogen). A line of best fit was added to this graph using a
spreadsheet program to enable the intercept (the time when the
temperature of the sample reached 77 K) to be calculated exactly.
The original results from the gas volume meter were then plot-
ted against time, and a line of best fit added using the spreadsheet
program (see Table (6.2) for an example of this graph). When the
intercept from the first graph was substituted into the equation of
the second line of best fit, the reading on the gas volume meter at
the time when the temperature of the sample was 77 K was found.
The reading from the gas volume meter at t = 0 was then subtracted
from this to find the actual volume of N2 evolved.
This volume was then converted into the number of moles of N2
using the theory laid out in the Number of Moles of a Gas section
and into a number of grams.
Using the average value of latent heat of vaporization of N2 cal-
culated earlier this was converted into a number of Joules. Dividing
by the change in temperature (i.e. 293 K – 77 K = 216 K or 87 K
– 77 K = 10 K) gave a figure in Joules per Kelvin. Using the mass of
the sample and the Mr, the number of moles of the sample was cal-
culated and then the specific heat capacity in Joules per Kelvin per
mole (JK−1mol−1) was calculated.
Calculation of Results Over Temperature Range 86 K  77 K
The background reading observed during the first 600 seconds was
extrapolated to cover the time period of the rest of the experiment.
The difference between the actual volume and the extrapolated vol-
ume was then used in the same way as the change in volume was used
in the Calculation of Results Over Temperature Range 293 K → 77 K.
266 • Physics Lab Experiments

Observations and Results

Latent Heat of Vaporization Results

Table 6.1: Latent heat of vaporization results

The average Lvap was found to be 249 kJkg−1 with a propagated error
calculated to be ± 4 kJkg−1.
Mass, Mr, Volume, and Density of Samples

Table 6.2: Sample measurements

 Common Experiments • 267

4.3 Specific Heat Capacity Over Temperature

Range 298 K  77 K

Table 6.3: Specific heat capacity results

Figure 6.25: Example of a graph used to calculate the time when the background level was
reached. This graph refers to the first result for copper over the range 293 K to 77 K. The
horizontal line shows the background rate and the curve is a line of best fit added to the
points showing the rate at which the nitrogen was evolved during the experiment.
The equation is the equation of the line of best fit.
268 • Physics Lab Experiments

Figure 6.26: Example of a graph used to calculate the volume of nitrogen evolved in the time
taken for the rate of production of nitrogen to reach background levels. This graph again
refers to the first result for copper over the temperature range 293 K to 77 K. The curve is the
line of best fit added to the points showing the reading on the gas volume meter. The equa-
tion is the equation of the line of best fit.

Note on validity of equations of lines of best fit: the choice of the

form of these equations was based upon finding an equation that fit-
ted the points well and not upon any assumed physical dependence
of one variable on another. This method is merely an improved ver-
sion of drawing a straight line up from the x-axis to a point on a
hand-drawn line of best fit and the drawing a line across to the y-axis
to take a result. Clearly the equations of the lines of best fit are not
valid outside the range over which the experiment was performed,
but within the range they provide the most accurate method of eval-
uating results from the graphs.
Specific Heat Capacity Over Temperature Range 87 K 
77 K

Table 6.4: Specific heat capacity results

 Common Experiments • 269

Accepted Results

Latent Heat of Vaporization

The accepted value of the latent heat of vaporization of N2, Lvap,
is 199 kJkg−1.

Specific Heat Capacity

Table 6.5: Specific heat capacity results

Density

Table 6.6: Table of densities

270 • Physics Lab Experiments

Discussion

Latent Heat of Vaporization of N2

It was found that Lvap=249 ±4 kJkg−1. In comparison with the
accepted value of 199 kJkg−1, the experimental value appears to have
some significant non-statistical error since the accepted value lies
well outside the statistical error range.
By examining the experimental results it can be seen that “too
much energy is needed to boil off each kg of nitrogen”. The energy is
probably being lost in the heater. When the power supply to the heater
was turned off the volume of nitrogen evolved was no longer measured.
At this point some of the energy calculated to have been transferred to
the liquid nitrogen had not done so since the heater was sill hot.
The experiment could be modified so as volume readings were
continued after the heater was turned off until the rate of produc-
tion of the gas reduced to background levels. This way all the energy
calculated to have gone into the heater would have been transferred
to the liquid nitrogen. This should produce a more accurate value
for the latent heat of vaporization.
Specific Heat Capacity Over Temperature Range 298 K
 77 K
The results calculated for the aluminium, the second copper read-
ing, the first carbon, reading and the lead readings were all signifi-
cantly too high when compared with the accepted results. (See the
Specific Heat Capacity Over Temperature Range 298 K → 77 K and
the Specific Heat Capactiy sections). The accepted results give the
specific heat capacity at a particular temperature and are not directly
comparable to those obtained experimentally over a large temperature
range. But, since specific heat capacity is a single valued function of T
which increases as temperature increases, it is expected that the spe-
cific heat capacity over a large temperature range should fall between
the accepted specific heat capacities at the bottom and the top of the
temperature range. For these results this clearly isn’t the case.
The other results (i.e., the first copper, the zinc, the second
carbon, and the salt) fall within the accepted results given for 50 K
and 298 K – this is to be expected since the experiment was carried
out between these temperatures.
 Common Experiments • 271

The fact that the experimental results are consistently too

high in comparison with the accepted values is explained by the
latent heat of vaporization for liquid nitrogen calculated earlier in
the experiment being significantly too high. If this had been more
accurate then it would have reduced the results for specific heat
capacity, making them in much better agreement with the accepted
values.
Another possible factor contributing the high values obtained
for the specific heat capacities was the variation in the background
count between the beginning and end of the experiment. It was
assumed to be constant when testing over the range 293 K  77
K. This inaccuracy was rectified by an improved method for testing
over the range 87 K  77 K as outlined in the Specific Heat Capac-
ity Over Temperature Range 87 K → 77 K section.
Specific Heat Capacity Over Temperature Range 87 K 
77 K
The general trend shown by the Debye model that the heat
capacity decreases as temperature decreases can be observed from
these results in comparison with those obtained over the tempera-
ture range from 293 K to 77 K. The specific heat capacity over the
range 293 K to 77 K is much higher than that over the range 87 K
to 77 K.
The results are still too high in comparison with the accepted
results. The error range is very high, and nearly as large as the result
itself for carbon, since the volume of nitrogen evolved is very small.
(Since the sample is at 87 K when it is put into the liquid nitrogen at
77 K, there is much less heat energy that can be used to vaporize the
liquid nitrogen than when the sample is originally at 293 K). Again
the reason for the results being too high is the very high latent heat
of vaporization obtained earlier.
Density
The density of the samples was calculated, so as this could be
compared to the accepted density of the pure material (for which the
accepted results are quoted). There is a small difference between the
calculated and accepted densities, but this should not be significant
enough to account for the variation seen between the experimental
and accepted values of the specific heat capacity.
272 • Physics Lab Experiments

Conclusion

The latent heat of vaporization of liquid nitrogen was found to

be 249 ± 4 kJkg−1, significantly higher than the accepted value of 199
kJkg−1. The specific heat capacity over the temperature range from
293 K to 77 K was generally found to be higher than the accepted
value, but this is explained by the latent heat of vaporization being
significantly higher than the accepted value. When the samples were
cooled to liquid argon temperature, the specific heat capacities still
remain high, but at one extreme of the error range the results are
comparable with accepted values.
As the temperature range drops, the specific heat capacity
decreases qualitatively inline with Debye theory and the accepted
values.

Bibliography

[1] MacDonald, D. K. C. (1964) Introductory Statistical

Mechanics for Physicists, John Wiley and Sons (USA)
[2] Reif, F (1965) Statistical and Thermal Physics,
McGraw-Hill Book Company (London)
[3] Baierlein, Ralph (1999) Thermal Physics, Cambridge
University Press (Cambridge)
[4] Website: Archives in London and the M25 Area,
www.aim25.ac.uk
[5] Website: Glasgow University Chemistry Department,
http://www.chem.gla.ac.uk/ alanc/dept/black.htm
[6] Muncaster, Roger (1993) Physics, Stanley Thornes
(Publishers) Ltd. (Cheltenham)
[7] Lyons, Louis (1991) A Practical Guide to Data Analy-
sis for Physical Science Students, Cambridge Univer-
sity Press (Cambridge)
[8] Kaye and Laby (1973) Tables of Physical and Chemi-
cal Constants, Longman
[9] Dewar, J. Proc. Royal Society A76 325, (1905)
 Common Experiments • 273

REFERENCES

[1] http://www.cleapss.org.uk
[2] Mythbusters TV Program, Discovery Channel, Epi-
sode 63 (2006).
[3] British Compressed Gases Association Code of Prac-
tice CP30 “The Safe Use of Liquid Nitrogen Dewars
up to 50 liters’.
[4] http://www.esrl.noaa.gov/gmd/ccgg/trends/global.html.
[5] Bureau International des Poids et Mesures: www.
bipm.org/utils/common/pdf/its-90/SUPChapter4.pdf.
[6] M. M. J. French and Michael Hibbert. Making liquid
oxygen. Phys. Ed., 45:221, 2010.
[7] M. M. J. French. Mobile phone faraday cage. Phys.
Ed., 46:290, 2011.
[8] J. G. Ziegler and N. B. Nichols. Optimum settings for
automatic controllers. Transactions of the ASME.,
64:759, 1942.
[9] A. S. McCormack and K. R. Godfrey. Rule-based
autotuning based on frequency domain identification.
IEEE Transactions on Control Systems Technology,
6:1, 1998.
[10] http://www.amsterchem.com/onlinehelp.
php?page=welcome.htm.
[11] M. Vannoni and S. Straulino. Low-cost accelerom-
eters for physics experiments. Eur. J. Phys., 28:781,
2007.
[12] F. Rooney and W. Somers. Using the wiimote in intro-
ductory physics experiments. TCNJ J. Stud. Schol-
arsh., 12:1, 2010.
[13] M. D. Wheeler. Physics experiments with nintendo
wii controllers. Phys. Ed., 46:57, 2011.
[14] http://wiimotephysics.codeplex.com/.
[15] http://pdg.lbl.gov/2012/tables/rpp2012-sum-quarks.
pdf.
[16] J. A. Nelder and R. Mead. A simplex method for func-
tion minimization. Computer Journal, 7:308, 1965.
[17] http://www.openoffice.org.
274 • Physics Lab Experiments

[18] http://www.miktex.org.
[19] http://www.winedt.com.
[20] http://www.irfanview.com/.
[21] M. M. J. French et al. Changing fonts in education:
How the benefits vary with ability and dyslexia. The
Journal of Educational Research, 2013.
[22] http://www.prezi.com.
INDEX

A Beta source, 240

AA battery, 45 Binomial distribution, 225, 227–228, 234
Acceleration Black, Joseph, 252
with one light gate, 71 Bluetooth radio adapter, 108
with two light gates, 71 BNC connector, 69
Accumulated error, 85 Boltzmann constant, 80
Accuracy and precision, 115–116 Bourdon tube, 77–78
Alpha decay, 231
Ammeters, 50–51 C
Ampere’s law, 63 Calibration, 66, 78, 83, 107, 205,
Analogue oscilloscope, 69 graph, 230–232, 234, 237, 244
Angular velocity, 218 Capacitance thermometry, 83
Anode-screen distance, 218–219, 221 Carbon dioxide, phase diagram for, 31
Appropriate significant figures in final Carbon-glass thermometers, 82
answers and errors, 134 Cathode ray tube, 220
Argon gas, 36 CBCONF32 program, 98–99, 102
phase diagram for, 37 CCD sensors, 105
Atomic force microscope (AFM), 90 Center-zero analog meter, 119
Axial electrode, 43 Character Map program, 190
Characteristics of laboratory, 2–4
B Charge:mass ratio of electron, 219, 223
Background count rate, 235 Chemical hazard symbols, 11
Background noise, 58 Chi squared test, 229, 234
Balances, 49–50 Chromatic aberations, 87–88
Bayard-Alpert gauge, 78 Clarity of arguments, 7
BCGA Code of Practice, 29 Clausius-Clapeyron equation, 253
Becquerel, Antoine Henri, 225 CLEAPSS, 9
Beta minus decay, 231 Closed cycle cryogen systems, 38
Beta plus decay, 231 Cold cathode gauges, 78
276 • Physics Lab Experiments

Colored chemical thermometers, 79 MATLAB, 147–151

Common experiments mean of sample, 120
Foucault’s method, 210–215 measurement distributions,
measuring acceleration due to 123–125
gravity, 202–208 measurement errors, 117
measuring the speed of light, Nelder-mead method, 160–164
208–209 plotting appropriate graphs,
optical fiber method, 209–210 134–135
Component frequency, 58 products and quotients, 131–132
Component/sample resistance, 53 random error, 117
Computer data logging, 97–105 regression line of Y on X, 151–154
Computers, 92–96 resolution, 118
Condensed matter laboratory, 3 response time, 119–120
Contact configuration, 55 sensitivity, 119
Controller output, 84–85 standard error on single
Correction factor, 219 measurement, 126
CRT oscilloscope, 65–66 standard error on the mean,
Cryogen free cooling, 25, 37–38 126–127
Cryogen free systems, 38 steepest descent, 154–155
Curie, 226 straight line graphs, 136
Cryogenic, 25 systematic error, 117
gloves, 27 tolerance, 118–119
uncertainty in gradients and
D intercepts of graphs, 165–168
Data analysis and errors uncertainty, 118
accuracy and precision, 115–116 variance and standard deviation,
appropriate significant figures in 121–122
final answers and errors, 134 window functions, 171–172
distribution testing, 164–165 zero error, 117–118
drawing lines of best fit and Data harvest, 74–77
finding gradients, 135 Data Harvest EasySense, 74
error bars on graphs, 135–136 graph of light level against time, 76
errors estimation, 127–128 Data Harvest QAdvanced Datalogger, 75
errors propagation, 129 Data loggers, 108
excel 2003 and prior versions, Dead time measurement, 235
138–140 Debye temperature, 257
excel 2010, 140–144 Debye theory, 252–255, 272
Fourier transform, 168–171 Delphi, 97–98
Gaussian error distribution, Derivative gain constant, 85
128–129 Dewar
Gauss-Newton, 155–158 capacity, 30
general functions, 133–134 constant, 263
graph drawing software, 137–138 flask, 26, 252–252, 261
Kaleidagraph, 144–147 Dewar, James, 26, 252
least squares fitting, 151 Diaphragm pump, 22
linear error propagation, 130 Difference frequency, 58
log graphs, 137 Differential amplifier, 63
 Index • 277

Diffraction/interference patterns, 105 charge mass ratio of an electron,

Diffusion pump, 23 216–224
Digital diagram, 210
balances, 49 finding moment of inertia,
calipers, 48 205–208
oscilloscope, 65, 69 Foucault’s method, 210–215
Distribution investigating radioactivity, 225–251
curve, 125 latent heat of vaporization of
function, 125 liquid nitrogen, 251–274
testing, 164–165 measuring speed of light, 208
Drawing curved lines, 135 optical fiber method, 209
Drawing lines of best fit and finding rotating mirror changes angle, 211
gradients, 135 Excel 2003 and prior versions, 138–140
Dry ice, 25, 30–31 Excel 2010, 140–144
Dulong and Petit method, 253 Excitation current, 59
Experimental methods and problems, 2
E Experimental physics, 2
Easel, 111 Experimentalist, 1
Einstein temperature, 256–257 Exponential attenuation, 225, 236–237,
Electrical measurements, reducing 241, 245
noise, 61–64 results, 248
Electricity safe, 12–13 Extra high tension (EHT) supplies,
Electromagnetic radiation, 79–80, 110 13–14
Electromagnets, 89
Electron diffraction pattern from gold, F
90 Faraday’s law, 63
Electronic thermometry, 83 Fast Fourier transform (FFT), 168
EMF power supply, 13–14 complex, 169
Emission/absorption spectra, 105 frequency, 170–171
Energy dissipation, 259 magnitude, 169–171
Environmental scanning electron spectrum, 170
microscope, 91 Fixed value error bars, 142–143
Equations, 185–189 Fourier analysis window, 169
LaTeX, 186, 190 Fourier transform, 168–171
LaTeX special characters, 190–191
Microsoft Word, 185–187, 190 G
OpenOffice Writer, 188–189 Gadolinium-Aluminium-Arsenic
Equipment and materials, 2 (GaAlAs) semiconductor, 83
Error bar Gamma
on graphs, 135–136 particles, 43
settings, 146 sources, 42, 241, 245
variables, 145–146 rays, 43, 231
Errors estimation, 127–128 Gas cylinders, safety, 15–19
Errors propagation, 129 bracket and belt anchoring a
Example reports cylinder, 15
anode - cathode potential colors, 19
differences, 216 gas regulator, 17–18
278 • Physics Lab Experiments

key, 18 Interference/diffraction pattern, 92

spindle, 18–19 Internal resistance, 55
store gas, 17 Inverse square law, 43, 235–236, 240
transporting cart, 16–17 Inverted magnetron, 78
Gas expansion factor, 30 Ionization pressure gauges, 78
Gas leaking, 18 Ionizing radiation regulations 1999, 41
Gas pressure, 78 Ionizing radiation, 40–44
Gas thermometry, 82 IR thermometers, 79–80
Gaussian distribution, 130 Isotopes and types of decay, 231
Gaussian error distribution, 128–129
Gauss-Newton, 151, 155–159 K
algorithm, 158 Kaplan, 236, 244–246
GPIB bus, 102 Kaye and Laby tables, 86–87
GPIB interface, 98, 101–102 of physical and chemical
GPIB PCI card, 97 constants, 87
Geiger counter, 41 Kaleidagraph, 144–147
Gifford-McMahon (GM) coolers, 38 adding error bars in, 147
Geiger-Muller tube, 43, 229, 231, 233, adding trendline, 146
235–237, 244, 246 fit parameters box, 146
Globally harmonized system (GHS) plotting a graph, 145
of classification and labeling of selecting data for a graph, 145
chemicals, 10 Kirchoff’s first law, 56
Graph drawing software, 137–138
L
H Lab book, 174–175
Hand pumps and vacu-vins, 22 LabView, 97, 101–105
Hanning, 172 ‘GPIB read’ component in, 103
Horizontal error bars, 135 within ‘while’ loop, 105
Hazard information, 10 Lasers, 38–40
Hazard symbols, 10–12 classification, 38–40, 42
chemical, 11 goggles, 40
European descriptions, 10 retina damage and blindness, 38
non-chemical, 11–12 Latent heat
Heat capacity, 255 history, 252
Helical rotations, 218 importance of, 253–254
Helium liquefier, 32 of vaporization, 25–26, 32–33,
Helium-3, 33–35 251–252, 254, 261–262, 265–
temperature and vapor 266, 269–272
pressures, 35 Latex/nitrile gloves, 42
Helium-4, 32–34 Levenberg-Marquardt
temperature and vapor pressures, 34 algorithm, 160
High temperatures, 84 method, 158
Hot cathode gauges, 78 Least squares fitting, 151
Leidenfrost effect, 27
I Light emitting diode (LED)
Instantaneous error, 85 transmitter, 209
Integral gain constant, 85 Light gates, 70
 Index • 279

acceleration with one light gate, 71 other researcher’s data, 107–108

acceleration with two light gates, 71 PASCO science workshop
velocity with one light gate, 70 interface, 72–74
velocity with two light gates, 71 PID and temperature control,
Linear attenuation coefficient, 236, 244 84–86
Linear error propagation, 130 pressure gauges, 77–78
Log graphs, 137 resistor color codes, 52
Liquid cryogens, 25, 38 SI prefixes, 47
Liquid helium, 32–33 SI units, 46
Liquid nitrogen, 25–30 signal generator, 64
asphyxiation, 28–29 spectroscopes, 92
Dewars, 28–29 still cameras and scanners,
goggles, 28 105–106
neoprene plastic bucket for temperature measurement, 78–79
storing, 27 traditional photographs, 109–111
safety precautions, 33 transmission electron microscope,
Liquid oxygen, 35–36 89–90
Lock-in amplifers, 58–61, 63 TTL (transistor-transistor logic),
Lock-in oscillator output, 56–57 64–65
Low temperatures, 82 velocity with one light gate, 70
velocity with two light gates, 71
M vernier calipers, 47–48
Making measurements video cameras, 106–107
acceleration with one light gate, 71 voltmeters, 51–52
acceleration with two light gates, 71 WII controllers, 108–109
ammeters, 50–51 MATLAB, 97, 100, 147–151
balances, 49–50 command window, 164
computer data logging, 97–105 function, 161
computers, 92–96 graph with vertical error bars, 150
constant current sources, 55–57 script, 162
data harvest, 74–77 Mean of sample, 120
digital calipers, 48 Mean square error, 130
further ideas on measurement of Measurement
resistance, 53–55 distributions, 123–125
further ideas on reducing noise in errors, 117
electrical measurements, 61–64 instruments, 2
high temperatures, 84 Mechanical gauges, 77–78
Kaye and Laby tables, 86–87 Mercury and alcohol thermometers, 79
light gates, 70 Micrometer, 48–49
lock-in amplifers, 58–61 Microscopes, 87–89
low temperatures, 82 using graticule, 106
micrometer, 48–49 Microsoft Bluetooth Stack drivers, 109
microscopes, 87–89 Microsoft Excel regression package,
multimeters, 52 166–168
near room temperature, 79–82 Microsoft Excel 2003 regression
oscilloscopes, 65–70 package output, 168
other microscopes, 90–92 Motion sensor, 73
280 • Physics Lab Experiments

Multimeters, 52–53 PID

Muller, Geiger, 225 algorithm, 84–85
and temperature control, 84–86
N gain constants, 86
NASA Grace mission, 50 Piezoelectric strain gauge, 78
National Instruments compact DAQ, Pirani gauge, 78
102 Planck’s constant, 80
Near room temperature, 79–82 Planck’s Law, 80
Nelder-Mead method, 160–164 Plugs, sockets and connectors, 92–96
example data, 163 Poisson approximations, 228, 234,
simplex method, 164 238–239, 244
Newton’s second law, 230 shape of, 234
Nintendo Wii games console Poisson distribution, 225, 244, 246
controllers, 108 Portable appliance test (PAT), 12
Nitrogen, phase diagram, 26 Positive temperature coefficient, 82
Non-chemical hazard symbols, 11 Posters, 192
Normal approximations, 228, 238, 244 example, 195–196
shape of, 234 Post graduate laboratory demonstrator, 4
Normal distribution, 128, 225, 234, 246 Potential difference, 9, 13–15, 44–45,
51–57, 60–69, 72, 78, 81, 89, 91,
O 118, 134, 136, 216–217, 219,
Op-amp output, 57 222, 223, 230–234, 237, 244,
Optical microscopes, 87–89 259, 262
Optics laboratory, 3 connectors, 14
Oscilloscopes, 65–70 Presentation, 7, 194–200
example set of slides, 197–199
P fonts and color, 197
Paper sizes, 193–194 Presenting results
OpenOffice Impress, 194 equations, 185–189
Microsoft PowerPoint 2003, 194 graphs, tables and diagrams,
tolerances, 194 179–183
PASCO Capstone, 72 lab book, 174–175
PASCO DataStudio, software, 72 paper sizes, 193–194
choosing a sensor in, 73 posters, 192
to measure the velocity in a presentations, 194–200
photogate, 74 referencing, 183–185
to produce a distance time graph, reports, 175
75 software, 175–176, 192–193, 200
PASCO science workshop interface, special characters, 189–192
72–74 structure, 177
box connected to photogate, 72 Pressure
PCI cards, 97 difference, 77
Penning gauge, 78 gauges, 77–78, 82
Permanent magnets, 89 units, 20–21
Phase sensitive multiplier, 58 Probability distribution function, 227
Phase-sensitive detection, 58 Product moment correlation
Pictograms, 10 coefficient, 152
 Index • 281

Products and quotients, 131–132 electrical, 12–13

Proportional gain gas cylinders, 15–19
constant, 85 hand pumps and vacu-vins, 22
results, 85 hazard symbols, 10–12
Proportional output, 85 helium-3, 33–35
Protactinium and Radon, 43 high voltages, 13–15
Pythagoras’ theorem, 130 ionizing radiation, 40–44
lasers, 38–40
Q liquid helium, 32–33
Quantum harmonic oscillators, 253 liquid nitrogen, 25–30
Quantum theory, 258 liquid oxygen, 35–36
pressure units, 20–21
R rotary pump, 22
Radioactive decay, 226–227 shoes, 10
Radioactive isotopes, 40 turbomolecular pump, 23
modern day uses, 227 vacuum pumps, 19–20
Radioactive particles, 41 Safety advice, 10
Radioactive sources, 43 Sagan, Carl, 197
Radioactivity, 225–231 Sample resistance, 54
Random error, 117 Scanning electron microscope (SEM),
Reference frequency, 58 90
Referencing, 183–185 Scanning tunnelling microscope
LaTeX, 184 (STM), 90–91
the Harvard system, 184 Science Workshop 500 Interface, 72
the numerical system, 183–184 Science Workshop Digital Sensors, 73
Regression line of Y on X, 151–154 Scientific content, 7
Ratchet mechanism, 48 Scientific literacy and style, 7–8
Resistance thermometry, 82, 84 Secondary electrons, 91
Resistor color codes, 51–52 Secondary ionization events, 43
Risk assessment, 9 Seebeck effect, 81
contamination of food, 10 Sensitivity, 119
Resolution, 118 Sensor Bar, 109
Response time, 119–120 SI prefixes, 46–47
Rotary backing pump, 23 SI units, 46
Rotary pump, 21–23 Signal generator, 64–65
Rotary vacuum pump, 25, 33 Simple four-wire mounting
R2 value, 154 configuration, 55
Rutherford, Ernest, 226 Software
Google Docs, 176
S LaTeX, 176, 192
Safety Microsoft Office suite of
argon, 36–37 programs, 176
cryogen free cooling, 37–38 Microsoft PowerPoint, 192
cryogenics, 25 Microsoft Word, 175, 181, 192
diaphragm pump, 22 MiKTeX, 176
diffusion pump, 23 OpenOffice Impress, 192
dry ice, 30–31 OpenOffice website, 176
282 • Physics Lab Experiments

OpenOffice Writer, 175, 181, 192 Thompson’s experiment, 223

WinEdt, 176 Thorium-232, 43
Special characters, 189–192 Tissue weighting factor, 41
Greek character, 189–190 Tolerance, 118–119
Special cryogenic gloves, 31 Traditional photographs, 109–111
Specific heat capacity, 252, 255–256, Transmission electron microscope,
258, 262, 264, 264, 268–271 89–90
history of, 253 Transmission microscopes, 87
importance of, 253–254 Transmitted electrons, 91
Spectroscopes, 92 TTL (transistor-transistor logic),
Spherical aberations, 88 64–65
SR830 lock-in amplifier, 98 Turbomolecular pump, 23–25
Standard error on single measurement, with rotary pump, 24
126
Standard error on the mean, 126–127 U
Stanford Research Systems (SRS) Uncertainty in gradients and intercepts
830, 59 of graphs, 118, 165–168
Steepest descent, 154–155 Undergraduate laboratories, 25, 110
Stefan-Boltzmann Law, 80 at University of Oxford, 3
Still cameras and scanners, 105–106 Undergraduate physics laboratory, 37
Straight line graphs, 136 calculations and error analysis, 5
Structure, presentation demonstrators’ responsibility, 5
abstract, 177 high voltages, 13–15
appendix, 178 lab demonstrator or lab
conclusion, 178 technician, 12
experimental methods, 178 demonstrating, 4–8
introduction, 177 electricity safe, 12–13
references, 178 Underground cavern, 114
results and discussion, 178 Uranium-238, 43
theory, 178
title, 177 V
Suede gardening gloves, 27, 31 Vacuum flask, 26
Systematic error, 116–117 Vacuum pumps, 19–20
plastic pressure vessels, 20
T Vacu-vin hand pump, 21
Tab delimited, 137 Variance and standard deviation,
Taylor expansion, 158 121–122
Teachers’ instructions, 9 Velocity
Temperamental equipment, 6 of the electron,
Temperature measurement, 78–79 with one light gate, 70
Theoreticians, 1 with two light gates, 71
Thermal equilibrium, 78 Vernier calipers, 47–48
Thermionic emission, 89 Vertical amplitude, 66
Thermocouples, 80, 83 Vertical gain, 66
Thermoelectric effect, 81 Video cameras, 106–107
Thick film resistors, 83 Voltage response, 58
Thin film resistors, 82 Voltmeters, 51–52
 Index • 283

W Y
“Wiimote” software, 108–109 Y error bars tab, 140
“position sensor” mode, 109
Window functions, 171–172 Z
Wii controllers, 108–109 Zero error, 117–118
Ziegler–Nichols method, 86
X Znamia Project, presentation, 198–199
X axis error bars, 143
X error bars tab, 140