Written by Shu Yu Yang Candidate number:

Wine glass acoustics

How does the frequency vary with the water amount in a singing wineglass?

Camilla Shu Yu Yang

1|Page
Written by Shu Yu Yang Candidate number:

Abstract
This Extended essay explores the following questions: How does the frequency vary with
the water amount in a singing wineglass? This paper will focus on how and why the
frequency changes as it does.

This investigation is approached through recording the frequencies for each height
interval. The frequency for each height interval will be recorded three times, and an
average was found. The PC software tool FFT Properties was used to calculate the
frequency of the measurements, and by using Microsoft Excel, we found the regression
line of our datas.

Using A.P. Frenchs paper on wineglass acoustics, we were able to prove mathematically
and through graphing tools how and why the frequency changed as it does. We also used
the program DataStudio to find the standard deviation of the data points compared to the
line we should have gotten when we plotted Frenchs equation.

By using Frenchs equation, we could explain that it was the shape of the wineglass that
determined the frequency change. The radius of the bowl at the water level and the
thickness of the glass wall contributed to a gradual decrease of the frequency at the
bottom part, while a steep decrease at the top.

2|Page
Written by Shu Yu Yang Candidate number:

Table of Contents
Abstract: .....................................................................................................................................1
Introduction ................................................................................................................................4
The theory:..................................................................................................................................5
Superposition: .........................................................................................................................5
Standing waves, nodes and antinodes ......................................................................................6
Slip and Stick effect: ...............................................................................................................7
How does the glass vibrate?.....................................................................................................7
Resonance frequency:..............................................................................................................8
Beating....................................................................................................................................9
Linear regression, Standard deviation and Correlation ...........................................................10
Relationship between the three operations..........................................................................11
A.P. Frenchs general formula ...............................................................................................12
The experiment .........................................................................................................................13
Variables:..............................................................................................................................13
Procedure and Setup..............................................................................................................14
Equipment:........................................................................................................................14
Setup.................................................................................................................................14
Data processing .....................................................................................................................16
Raw data ...........................................................................................................................16
Observations .....................................................................................................................16
Processed data.......................................................................................................................17
Data with mass as independent variable .............................................................................17
Processed data according to Frenchs general equation ......................................................19
Analysis ................................................................................................................................22
Conclusion................................................................................................................................23
Further investigation .................................................................................................................23
Evaluation:................................................................................................................................23
Appendix ..................................................................................................................................25
Tables of raw data: ................................................................................................................25
Bibliography ...............................................................................Feil! Bokmerke er ikke definert.

3|Page
Written by Shu Yu Yang Candidate number:

Introduction
The singing wine glass, also known as glass harp or the Harmonica, is a glass instrument
invented in 1714 by Richard Pockrich, and a modern version of the Harmonica is said to
be invented by Benjamin Franklin. The sound of this beautiful instrument is created when
energy is applied to the glass and the physical properties of the glass start to resonate.
The applied energy will then in this case be rubbing one or two wet fingers on the rim of
a glass, and the glass will then start to vibrate, at its natural frequency. However, when
energy is applied externally, this frequency will be called the resonate frequency.
Different tones can be obtained by varying the amount of water inside the glass. Wine
glasses are usually used, so that one can hold onto the stem while rubbing to minimize
the interference of the singing. So my research question will be:

How does the frequency vary with the water amount in a singing wineglass?

The approach I will be using to answer my research question will be to measure the
frequencies with different amount of water. Then I am going to find a relationship
between the frequency outcome and how we vary the water. A.P. French has written a
journal paper1 about this specific type of resonance use on wineglasses, he derived a
general formula for how the frequency could vary with the water amount in the glass, so
this paper will be used to confirm our experiment. The program FFT-properties2 is the
program used to obtain the records taken during the experiment as frequency
The rim
(Hz) values. By analyzing the different frequencies and
process them, we are able to find a relationship The bowl
between the frequency variation and amount of water
used.
The stem

The foot
3
Figure 1 - A wine glass

1
In Vino Veritas A study of wineglass acoustics A.P French, Am. J. Phys., Vol.51, No. 8, August 1983
2
http://www.dewresearch.com/fftp-main.html
3
http://www.webstaurantstore.com/libbey-3057-perception-11-oz-wine-glass-24-cs/libbey-3057-
perception-11-oz-wine-glass-24-cs.jpg

4|Page
Written by Shu Yu Yang Candidate number:

The Theory:

Superposition:4
When there is more than one wave travelling in the same medium, the sum of the wave
functions is the algebraic sum of the values of the wave function. This is called
superposition, and is the basic theory behind standing waves. So when waves travel
towards each other, instead of breaking the rhythm totally, they travel through each other.
There are two types of superposition, constructive, where the net sum of the waves gives
a positive displacement, and destructive, vice versa.

<- Constructive
superposistion

Destructive
superposition ->

Figure 2 Superposition

So the net displacement of the wave will be as the picture states:

[T.1]
Superposition is the basis for the understanding of standing waves, which will be relevant
to the observations.

4
Basic theory from Physics for Scientists and Engineers 6E by Serway and Jewett unless otherwise
stated

5|Page
Written by Shu Yu Yang Candidate number:

Standing waves, nodes and antinodes

We can analyze situations where two or more waves travel together as standing waves.
Then we can consider two waves that are travelling towards each other, one is then
travelling to the right (y1) and one is travelling to the left (y2):

= ( ) = (+ )

Then applying the principles of superposition and [T.1], then you get:

= + = (+ )+ ( )

By using the trigonometric identity [sin( ) = sin cos cos sin ], we get:

= 2 sin cos

This is the standing wave function. When sin = 0, then the amplitude of a standing
wave motion is at a minimum. These are called zero points, more known as nodes! This
is obtained when:

= , 2, 3

Substituting k with = , we get:

3
= , , =  = 1,2,3
2 4 2

These zero points can be easily observed during the experiment.

6|Page
Written by Shu Yu Yang Candidate number:

Slip and Stick effect5:

The slip and Stick effect is a frictional effect where two surfaces are dragged along each
other to stick together and slips with very short intervals of time. The main theory behind
it is that friction between two surfaces wants to hold them together, thus stick. But the
external force on acting from one surface on the other is continuously trying to pull the
two surfaces away, therefore slip. Then the kinetic friction coefficient becomes larger
than the static friction coefficient, and vice versa when the two surfaces are sticked
together. That is why it is called the stick-and-slip effect. This motion is not a simple
harmonic motion. Violin uses this technique to produce the sound, or when you drag your
fingers over a balloon, but mainly our singing wine glass uses this to create the sounds.
By using this technique, the frictional jerking will create vibrations within the glass wall.

How does the glass vibrate? 6

When the glass is affected by the stick-and-slip motion, it starts to move in a very special
way. Since glass is quite elastic, the rim starts to deform
into an elliptical shape. This is proved by A. P. French7
through his derived equation of the horizontal radial
displacement x:

( = )( )

Where the [] factor describes the elliptical

movement from the circular mean shape. In this case
Figure 3 - How the glass vibrates

when , , there will be diagonal nodal lines which

are indicated in red lines in Figure. 4.

5
http://www.ccmr.cornell.edu/education/ask/index.html?quid=1143
6
http://www.ccmr.cornell.edu/education/ask/index.html?quid=1143
7
In Vino Veritas page 689.

7|Page
Written by Shu Yu Yang Candidate number:

Figure 4 - A vibrating glass with water in action8

We can clearly see the antinodes on this picture. It is where the water has maximum and
minimum displacement, crest and though, these are the antinodes. However, the nodes
and antinodes are not fixed in one position, but moving along your finger.

Resonance frequency:
If a periodic force is applied to such a system, the amplitude of the resulting motion is
greatest when the frequency of the applied force is equal to one of the natural frequencies
of the system.9 This is often referred as resonance, and the resonance frequencies are the
frequencies where the system exhibits relatively large amplitudes.

Figure 5 - Amplitude - frequency graph10

Figure 6 - Resonance frequencies from 50mm, 3.measurement11

8
http://www.flickr.com/photos/euicho/273665203/in/photostream/
9
Serway, page 558
10
Images from Serway, p. 558, screen printed
11
Screen printed image from the FFT-program.

8|Page
Written by Shu Yu Yang Candidate number:

The resonance frequency is essential in this experiment because the sound the wine glass
makes when its rim is rubbed by a moistened finger is the resonance frequency sound of
the glass with the specific amount of water.

The resonance frequency is also the main theory behind how to break a glass with sound.
Shortly summarized, when a sound played off close to the glass matches the resonance
frequency and is loud enough, the glass will continuously vibrate until the elastic
properties of the glass can no longer handle the energy from the sound flowing into the
glass, and hence break.12

Beating
Beating is a result of the superposition of two waves that are slightly out of phase, this
will give a periodic variation in the amplitude. Adding the superposition of two wave
functions with different frequency, where x=0 will become:

( )

Figure 7 - Beating13

Fig. 9 (a): two waves out of phase, one blue (y 1) and one black (y2)

(b): the superposition of the two waves out of phase (y 1 + y2)

12
http://www.fas.harvard.edu/~scdiroff/lds/OscillationsWaves/ShatteringWineglass/ShatteringWineglass.html
13
Figure of beating from Serway, p.565

9|Page
Written by Shu Yu Yang Candidate number:

Using the trigonometric identity: cos + cos = 2 cos cos

and rearranging it, we get:

= 2 cos 2
2

Through the figure and the equation, one will hear a sound of periodically varying
intensity, amplitude due to the cosine factor in the equation.

The beat frequency will then be:

= | |

For example, the frequency outcome of the wineglass empty was 796.3Hz and 800.2Hz.
The average sound wave frequency will be 798.25 Hz, which will be the resultant sound
wave frequency. There will be a beat frequency of 3.9Hz, so the sound of 798.25Hz will
then be heard with an intensity maximum approximately four times per second.

The reason we get beating in our experiment is due to the microphone of the recorder is
placed beside the wineglass. When the wineglass vibrates due to the periodically sliding
finger on the rim, it will periodically move slightly away and towards the microphone as
seen in Figure 4. This will cause a periodical variation of the intensity of the sound
recorded, and thus the beating theory can be applied to our results. This is also the theory
behind why we will get two peaks when we run our data in the FFT. So when processing
the data, we can average the two peaks and get the overall frequency.

Linear regression14, Standard deviation15 and Correlation16

Linear regression is the operation where it finds the line of best fit for the data points
considered. This is done by minimizing the sum of the square of the vertical distance
from the result values to the trend line. This operation is used on Excel and DataStudio to
find our trend line.

14
http://en.wikipedia.org/wiki/Linear_regression
15
http://en.wikipedia.org/wiki/Standard_deviation
16
http://en.wikipedia.org/wiki/Correlation_and_dependence

10 | P a g e
Written by Shu Yu Yang Candidate number:

Standard deviation is a tool used in scientific measurements, statistics and probability to

find out how much the results are scattered about the trend line of the values. If the
results are close to the trend line, then the standard deviation will be low. Is it is a large
value, then the data points are more spread out. The standard deviation can be calculated
through:

= ([ )] =

stands for the standard deviation, is the x value, is the mean value, while is the
operation the average of the expected values.

The correlation coefficient shows us the relationship between the two variables
considered. Pearsons correlation coefficient deals with strength of the linear relationship
between the two variables considered, but cannot show every characteristic the data
points have. can be found:

([ ) ]
=

( ) represents the horizontal distance from the data point to the regression line,
while is the vertical distance.

Relationship between the three operations

The relationship between the regression line and the standard deviation is that the slope
constant in = + is directly related to the standard deviation and the correlation
coefficient , which will be explained soon:

This is what the programs automatically give us later in the analysis part.

11 | P a g e
Written by Shu Yu Yang Candidate number:

A.P. Frenchs general formula17

In A.P. Frenchs paper, he presented an equation for variation of frequency variation:
= frequency of empty glass
= frequency of partially filled glass
1+ 1
5 (Hz)
= a constant
= density of liquid
His general formula was intended to describe an ideal = density of glass
= radius of water
cylindrical glass, but in his paper, he found out that this
= glass thickness
formula works approximately for any types of glass. = distance from top of glass top
water

= effective height of glass
As we can see, the part is build up by different constants Table 1 - Explanation of the signs used in the

equation by A.P. French
representing different properties of his ideal cylindrical glass,

we can substitute this with . The 1 part is representing the variable of this

equation, the height of the water . So we can substitute the

= 1 +

So if the wineglass is ideal and cylindrical, the best fit line for plotting against

1 should give us a totally straight and linear line.

17
In Vino Veritas, page 692

12 | P a g e
Written by Shu Yu Yang Candidate number:

The experiment

Variables:
The glass from IKEA was the only glass used during the whole experiment, so the shape
of the glass will not change, and so would not the frequency of the empty glass change,
because there would not be any interference to the frequency outcome if there is not
water in the glass to disturb the vibrations. Due to the fact that waters volume change
with temperature, the water used to the experiments was put away for a while before
using it to get it to an approximately room temperature. The frequencies were found by
processing the records on FFT-properties. There is an input control on the sound recorder,
which can be adjusted so that sound within a certain radius can be recorded. Before the
actual record, the recorder will show how the device reacts to the different sounds that it
catches, so I turned down the input and waited for everything to be completely quite (for
the recorder) before I started recording.

Two different experiments are to be done; one where the mass of the
water in the wine glass is varied and the other one is the height measured
from the bottom upwards. To minimize the uncertainties, a weight with
0.1g accuracy was used for the weight and a piece of spaghetti where
every 10cm was marked was used as a ruler that could measure the height Figure 8 - Measuring
instrument,
of the water from the bottom of the glass. The reason of using spaghetti spaghetti

was to minimize the interference of the water level with the measuring
instrument.

Independent variables: the height of water in (mm)

Dependent variables: the frequency (Hz)

Controlled variables: the frequency of an empty wine glass, the shape of wineglass and
the water temperature (C)

13 | P a g e
Written by Shu Yu Yang Candidate number:

Procedure and Setup

Equipment:
a) Wineglass (IKEA product nr: 000.151.34)
b) Water
c) Pipette
d) Spaghetti straw
e) Sound recorder, EDIROL R09
f) Processing program, FFT Properties version 5.0 Figure 9 - The equipments

Setup
We are first going to measure the mass/volume and the frequency outcome, then the
heights with the frequency outcome. To measure the exact volume, we use a weight to
find the volume, because 1gram of water = 1ml of water, the weight has an accuracy of
0.01g. We used a pipette to slowly add drops of water to get the amount of water as
accurate as possible.

The intervals chosen to test out the frequency variation is 10g for each interval, the sound
played was recorded as a WAV file on an EDIROL R09 sound recorder. Then the FFT
program18 is used to analyse the sound waves. The analysis will come out in an Intensity
(amplitude)Frequency graph, where the two peaks can be found. Using the peak finder
in the program to find the intensity of the peak and the frequency the peak lies on.

Figure 10 - the two peaks of 60 mm height, 3. measurement

18
FFT properties version 5.0, trial

14 | P a g e
Written by Shu Yu Yang Candidate number:

The almost same procedure is done with the measurements based on heights; just that the
way to measure the heights is through a spaghetti straw. We needed the height of the
glass from the exact bottom, so measuring with a ruler outside the glass would give
extremely vague accuracy. So I chose to mark the spaghetti with intervals of 10mm each.
Then put the spaghetti in the glass and try to hold it as straight as possible, and fill the
glass until the marks were reached. Measurements were taken when the water surface Spaghetti
seen from below touched the black line that indicated the intervals, the meniscus of the
water on the spaghetti straw.

To be accurate, each interval of investigation is measured three times. Measure all the
intervals in the glass once, then start over from the bottom again, and do this three times Figure 11 - how
the measurements
for both the mass and the heights. The first measurement is done at the student council were taken

office at my school, because that room has very good sound proofed, so the sound
recording will have minimum disturbance, and this is done when everybody else at the
school was having classes, nobody was in the vestibule.

In the end the frequency of an empty glass is measured, but only once is necessary,
because it will always be the same, compared to when there is water.

Each measurement lasted around 12-15 seconds, because when analyzing with the FFT,
we analyze the sound over a period of time, which I chose to be 3 seconds, so that any
random interference can be avoided by analyzing the sound from different parts of the
record.

The glass was dried properly with paper before starting on another measurement, to avoid
as much uncertainty as possible.

15 | P a g e
Written by Shu Yu Yang Candidate number:

Data processing

Raw data19
Table 2 - Raw data with height as independent variable:

Height (mm) 0.5 mm peak1 (Hz) 0.05 Hz peak2 (Hz) 0.05 Hz

1. measurement
0 796.3 800.2
10 796.1 798.9
20 795.3 798.1
30 791.4 794.3
40 778.0 781.2
50 742.9 745.9
60 676.8 679.7
70 582.0 585.0
80 477.5 480.8
90 375.4 378.5
The frequencies were found by analysing the sound waves over a period of time, which I
chose to be approximately four seconds. The frequency was displayed with three
decimals; I rounded it up to one decimal. The effective height of the glass is 90mm, total
height

Observations
The further up the water was, the easier the vibration waves were to be seen. You could
see the waves close to the glass walls, and standing wave patterns can be
observed. They occur 90degree to each other like: The circle is the glass, it is
the cross ( + )that copies my moves with my fingers when playing. These
wave patterns are the nodes of this motion described in the theory section.

The sound varied in intensity periodically, so it sounded like it has a certain Figure 12 - How
the wave moves
beat, rhythm of playing off one pitch.

19
In total, three measurements were taken for each independent variable, they can be found in the
appendix

16 | P a g e
 Written by Shu Yu Yang Candidate number:

Processed data

Data with mass as independent variable

Average of the datas

Height Average of Av. Of both
(cm) Average of Peak 1 Unc. Average peak 2 Peak 2 Unc. both peaks peaks Unc.
0.5 cm peak 1 (Hz) (Hz) (Hz) (Hz) (Hz) (Hz)
0 796.3 0.1 800.2 0.1 798.3 2.0
10 796.0 0.1 799.2 0.3 797.6 1.6
20 795.4 0.5 798.4 0.3 796.9 1.5
30 791.5 0.2 794.7 0.4 793.1 1.6
40 778.2 0.6 781.6 0.5 779.9 1.7
50 744.4 1.6 747.8 1.9 746.1 1.7
60 676.8 1.6 680.2 1.7 678.5 1.7
70 582.6 2.4 586.1 2.2 584.4 1.8
80 476.7 3.3 479.7 2.9 478.2 1.5
90 376.1 0.8 379.4 0.9 377.7 1.7
Table 3 - Processed data with height as independent variable

The uncertainty of the height is estimated of the measuring instrument.

17 | P a g e
 Written by Shu Yu Yang Candidate number:

Graph of average of both peaks

Average of both peaks

810,0

760,0

710,0
average of both peaks:
Poly. (average of both peaks:)
660,0
Frequency (Hz)

610,0

560,0

510,0

460,0

410,0

y = 4E-07x5 - 6E-05x4 + 0,0019x3 - 0,0161x2 - 0,1186x + 798,46

360,0
0 10 20 30 40 50 60 70 80 90 100
Height (mm)

Graph 1 - Average of both peaks, height as independent variable

18 | P a g e
 Written by Shu Yu Yang Candidate number:

Processed data according to Frenchs general equation

According to A.P. French, if we plot against 1 in a graph, we should get a

straight line, so we first process the data:

(mm)

1 1
0.5mm (Hz) (Hz)
0 789.3 2.0 1.000 0.005 1.000 0.01 0.0000 Error
10 797.6 1.6 0.990 0.004 0.979 0.01 0.0001 6E-05
20 796.9 1.5 0.990 0.004 0.981 0.01 0.0023 4E-04
30 793.1 1.6 0.995 0.004 0.990 0.01 0.0118 1E-03
40 779.9 1.7 1.012 0.005 1.024 0.01 0.0373 3E-03
50 746.1 1.7 1.058 0.005 1.119 0.01 0.0911 5E-03
60 678.5 1.7 1.163 0.006 1.353 0.01 0.1890 8E-03
70 584.4 1.8 1.351 0.007 1.824 0.02 0.3501 1E-02
80 478.2 1.5 1.651 0.009 2.724 0.03 0.5973 2E-02
90 377.7 1.7 2.090 0.014 4.367 0.06 0.9568 2E-02
Table 4 - Processed data 2

is the height of the water from bottom of, is the frequency at the different heights,
is the frequency of the glass when it is empty, is the vertical distance from the top of
the glass to the water surface and is found by ( ), is the effective height of the
glass which is 91 mm, and represents uncertainty of.

Through the STDEVA[value1,value2] operation in Excel, it found us the standard

deviation for the x and y variables. = 1.1096294 and = 0.3232.

Uncertainty of is found through:

= +

where is the final uncertainty, is the uncertainty of the frequency with an empty
glass and is the uncertainty of frequency with different amount of water.

Uncertainty of is found through:

19 | P a g e
 Written by Shu Yu Yang Candidate number:

= 2



Uncertainty of 1 is found through:

= +


Then finding the uncertainty of 1 will be:

1

= 1 4

1

Then plotting the graph
5

4,5

3,5
y = 3,3983x + 0,8765 Serie1
Liner (Serie1)
3
Log. (Serie1)
Poly. (Serie1)
(f0/f)2

2,5

2
y = 2.7022x4 - 5.6239x3 + 5.4911x2 + 1.065x + 0.9828
1,5

0,5

0
0,0000 0,2000 0,4000 0,6000 0,8000 1,0000 1,2000
(1 - d/H*)4
2 4
Graph 2 - (f0/f) plotted against (1 - d/H*)
20 | P a g e
Written by Shu Yu Yang Candidate number:

Plotting the graph into DataStudio20 , we can use its regression functions to find the
standard deviation for the data points to the trend line:

Graph 3 - Data plotted into DataStudio, everything is in norwegian

The equations of the two regression lines can be found in Graph 2.

The standard deviation for the linear best fit is 0.149, while it is 0.00995 for a polynomial
regression.

20
http://www.pasco.com/datastudio/

21 | P a g e
Written by Shu Yu Yang Candidate number:

Analysis
One observation from the results is that the frequencies of the wineglass with water at the
bottom decreased very slowly, this is because the bottom part of the wineglass has the
stem to reduce the distance the glass can vibrate. The wavelength gets smaller, and

according to the equation = , the frequency will become large the smaller the

wavelength. There is more space for the glass walls to vibrate the further up we go, so the
frequency will decrease. We get the space from the shape of the glass, the radius of the
top is way bigger than at the bottom, and the glass walls are thinner at the top. This will
give the glass a more elastically property at the top, and therefore make the walls to
vibrate more at the top. This can be proved mathematically through Frenchs general
equation.

We plotted our results in Frenchs equation and drew the graph of it as we did in Graph 2.
According to French, an ideal wineglass should have the data points that could have a
perfect linear trend line, but the linear trend line we drew did not fit our results perfectly.
Our data points showed a slightly curved trend line, as the blue line in Graph 2 showed,
this means that its gradient increases with the x-values. This can be explained when we
look at Frenchs equation21

1+ 1
5

We only need to look at and which are the radius of the glass and the thickness of the
glass at the water level. The density of the glass and the liquid is constant, since we made
sure that the temperature of the water used was approximately the same. is just a
constant he used. In our case, only R and a was varying, the further up the water level is
the bigger area it will have, thus increasing due the shape of the wineglass as seen in
Figure 1. The thickness of the glass will decrease the further up the glass walls, again

because of the shape. This will give an increasing value the more water we have, and

thus give the increasing gradient of the trend line, and explains the curved shape of it.

21
In Vino Veritas, page 692

22 | P a g e
Written by Shu Yu Yang Candidate number:

We have as well found how much the linear line does not fit the general equation of
Frenchs, with a standard deviation of 0.149. This is not a big number, but big enough to
show that other varying factors are involved in this experiment, like the thickness and
radius of the glass at the water level.

Conclusion
We have now investigated how the frequency varies of a singing wine glass, and how the
singing wine glass actually works as well as why the frequency decreases as it does. The
results showed that the frequency decreases the more water we add up in the wine glass,
however it decreases the most at the upper part of the wineglass, while it nearly does not
decrease at the bottom. This is definitely dependent on the shape of the wineglass tested.
Because of its shape, the frequency decrease did not fit the general equation of A.P.
French. We found out that it was the radius of the glass and the thickness of the glass
walls at the different water levels that contributed to a slightly curve when we tried to

plot against 1 . This explains how and why the frequency decreases the way

it does.

Further investigation
Original intentions were to test out different liquid types with different density, if that
will effect this relation or not. Say using yogurt will maybe give different results than
using water. The temperature of the liquid may also affect the results due to different
density of water at different temperatures, so trying out hot and cold water will also be
very interesting. Testing out the frequency change with different types of wine would
also be fascinating, like white and red wine, old and new wine. Different liquor and
alcohol can also be tried out, since, after all, this is a glass specifically made for drinking
alcohol!

Evaluation:
This experiment has many flaws; how you rub the rim is a great limitation to this
investigation. If you rub it too hard, the sound would not be coherent, if you rub it too

23 | P a g e
Written by Shu Yu Yang Candidate number:

soft, then you would not get enough of the stick effect to create a clear sound. Every time
I rubbed the glass, there were always some differences, so these uncertainties will
definitely affect the frequency outcome. The frequency is found by analysing the sound
over a period of time and use the peak of intensity, if the time interval was bigger, the
frequency outcomes would be more accurate. However, when measuring the sound, there
was only one microphone placed on one side of the wineglass, which produced a beating,
so if we have placed several microphones around the wineglass and recorded with all of
them, we might eliminate the beating.

Even though the input of the EDIROL was controlled to minimize the interference, there
may be interference from my side, rubbing, or random noises, tried to control by
analyzing the sound differently. Breathing was sometimes even noticed by the EDIROL.

When the water level was at the top, even analyzing with 4sec time period was too big.
There was too much interference of the water that was climbing on to the rim and down
the glass walls.

I held the glass stem while rubbing, this might have some impact on the results, and so if
the glass fastened on the foot to the table or anything, some small interference could be
cancelled. Of course, if a set up that allowed a robot to rub the rim and, but I have done
the best out of it already.

24 | P a g e
Written by Shu Yu Yang Candidate number:

Appendix

Tables of raw data:

Height (mm)

2. measurement Peak 1 (Hz) Peak 2 (Hz)

10 796.1 799.4
20 796.0 798.7
30 791.7 794.6
40 778.9 782.3
50 746.1 749.6
60 675.1 678.8
70 585.4 588.9
80 479.7 482.0
90 377.0 380.3
3. measurement
10 795.9 799.4
20 795.0 798.5
30 791.4 795.1
40 777.7 781.4
50 744.2 748.0
60 678.4 682.2
70 580.5 584.5
80 473.0 476.2
90 375.8 379.4

25 | P a g e
Written by Shu Yu Yang Candidate number:

Bibliography
Physics for Scientists and Engineers by Raymond A. Serway and John W. Jewett, Brooks Cole;
6 edition (July 21, 2003)

In Vino Veritas A study of wineglass acoustics A.P French, Am. J. Phys., Vol.51, No.
8, August 1983

Digital sources:

1) http://www.webstaurantstore.com/libbey-3057-perception-11-oz-wine-glass-24-
cs/libbey-3057-perception-11-oz-wine-glass-24-cs.jpg
2) http://www.ccmr.cornell.edu/education/ask/index.html?quid=1143

3) http://www.fas.harvard.edu/~scdiroff/lds/OscillationsWaves/ShatteringWineglass/
ShatteringWineglass.html

4) http://www.dewresearch.com/fftp-main.html

5) http://www.flickr.com/photos/euicho/273665203/in/photostream/

6) http://en.wikipedia.org/wiki/Linear_regression

7) http://en.wikipedia.org/wiki/Standard_deviation

8) http://en.wikipedia.org/wiki/Correlation_and_dependence

9) http://www.pasco.com/datastudio/

List of figures:

Figure 2 - A wine glass ....................................................................................................... 4

Figure 3 - Superposition...................................................................................................... 5
Figure 4 - How the glass vibrates ....................................................................................... 7
Figure 5 - A vibrating glass with water in action................................................................ 8
Figure 6 - Amplitude - frequency graph ............................................................................. 8
Figure 7 - Resonance frequencies from 50mm, 3.measurement......................................... 8
Figure 8 - Beating ............................................................................................................... 9
Figure 9 - Measuring instrument, spaghetti ...................................................................... 13
Figure 11 - The two peaks of 60 mm height, 3. measurement.......................................... 14

26 | P a g e
Written by Shu Yu Yang Candidate number:

Figure 10 - The equipments .............................................................................................. 14

Figure 12 - How the measurements were taken................................................................ 15
Figure 13 - How the wave moves ..................................................................................... 16

List of Graphs:

Graph 1 - Average of both peaks, height as independent variable ................................... 18

Graph 2 - (f0/f)2 plotted against (1 - d/H*)4 ...................................................................... 20
Graph 3 - Data plotted into DataStudio, everything is in norwegian................................ 21

List of tables:

Table 1 - Explanation of the signs used in the equation by A.P. French .......................... 12
Table 2 - Raw data with height as independent variable: ................................................. 16
Table 3 - Processed data with height as independent variable.......................................... 17
Table 4 - Processed data 2 ................................................................................................ 19

27 | P a g e