The relationship between the height of water level in a glass

and the frequency of the emitted sound

Introduction, aim and background information

On my last trip to Budapest, Hungary, I watched a street musician play an instrument

called the glass xylophone. A glass xylophone is a musical instrument made of upright

glasses that are filled with water to different heights. There are many ways to make a

glass xylophone, simpler instruments consist of just a few glasses, while more complex

ones are carefully designed and can include over fifteen glasses. Observing the street

musician play the glass xylophone made me curious about the Physics behind this

instrument, so for my internal assessment I decided to investigate how a glass

xylophone works, more specifically, why different glasses produce different pitches

depending on the height of the water in the glass. Pitch in music simply means how high

or low a sound is ("Pitch"). When studying waves, I remember learning that higher pitch

means that the sound wave has a higher frequency, while a lower pitch means lower

frequency. Based on what I know so far, I decided on the following research question

for my investigation:

How does the height of the water level in a glass affect the frequency of the sound

emitted by the glass when the side of the glass is hit with a spoon?

When I started researching glass xylophones, I found many more articles about singing

glasses, also called glass harps. In contrast to a glass xylophone, when playing a glass

1
harp one has to rub the glass rim with a finger to produce sound (Nose 4). I also

discovered that it does not matter whether we rub our wet fingers on the rim of the glass

or hit the side of the glass with a metal spoon, because both methods create vibrations

in the glass and produce the same frequency, so in my work I will also refer to articles

about singing glasses ("Singing Glasses"). When energy is transferred to the glass by

rubbing its rim or by hitting its side with a spoon, the glass starts to vibrate. In turn, this

causes the air molecules to vibrate with the same frequency, producing the sound that

we hear (Nose 10). As we add more water, it becomes more difficult for the glass wall to

vibrate due to the added mass and as a result, vibrates at lower frequencies, producing

a lower pitch (Lee 1). Moreover, I also found that the relationship between the height of

the water level and the frequency of the sound is not linear (Lee 3, Nose 11). Hence my

hypothesis is that as the water level in the glass increases, the frequency of the sound

produced by the vibrating glass will decrease at an increasing rate, producing a

nonlinear curve.

Planning

I. Variables

1. Independent variable: Height of the water level in the glass (cm)

2. Dependent variable: Frequency of the emitted sound wave (Hz)

3. Control variables

i. Glass used

● Reason: glass shape and size influence the vibrations

produced by the glass wall, hence affect the frequency.

2
 ● Method: I will use the same glass throughout the experiment.

ii. Glass cross-sectional area

● Reason: the volume of the water might influence the

frequency. If the glass has varying cross-sectional area,

every time we add water, the volume of the water added will

be different.

● Method: I will use a glass that has the same cross-sectional

area along its entire height.

iii. Liquid used in the glasses

● Reason: different liquids have different densities. Larger

density makes it more difficult for the glass to vibrate, hence

affects frequency.

● Method: I will use water throughout the experiment.

iv. Air temperature

● Reason: the speed of sound depends on the temperature of

the medium it travels in. Since frequency is related to wave

speed, temperature should be the same throughout the

experiment

● Method: I will carry out the experiment indoors so that

temperature can be kept constant.

3
II. Apparatus

Figure 1.​ Setup of apparatus.

III. Method

1. Place smartphone approximately 10 cm from the empty glass so that its

microphone is pointing towards the glass, open the Google Science Journal App

and turn on the Pitch sensor

2. Start the Pitch sensor

3. Hit the side of the glass with the spoon

4. Stop the Pitch sensor

5. Record maximum frequency displayed on the screen

6. Repeat steps 2 to 5 four more times

7. Increase water level by 2.0 cm

8. Repeat steps 2 to 7 until the water level reaches 12.0 cm

4
IV. Safety, ethical and environmental considerations

In general, this is a safe experiment. Since I am working with water, I will make sure

that when I am recording the frequencies, the smartphone is not connected to a

charger. I will also take care with the glass and ther jar, since these can break and

potentially cause injury. There are no animals or people involved in this experiment, so

there are no major ethical considerations. In addition, the experiment does not harm

the environment.

Data collection and analysis

I. Raw data

Uncertainties: in the table below, the uncertainty in ​H​ is ± 0.2 cm, because even

though I used a ruler with a millimeter scale division to measure the height of the

water level, it was somewhat difficult to read this measurement, so I chose a

larger uncertainty to be on the safe side. The uncertainty in ​f​ is ± 0.1 Hz,

because the Google Science Journal App shows the measured frequency to one

decimal place.

Table 1.​ Raw data collected from the experiment.

Height ​H /​ cm 0 2.0 4.0 6.0 8.0 10.0 12.0

ΔH = ± 0.2 cm

Trial 1 1156.3 1149.1 1139.7 1115.7 1030.2 908.7 709.1

Frequency
f​ / Hz Trial 2 1157.1 1148.3 1139.2 1116.0 1028.9 908.9 709.6

Δf = ± 0.1 Hz Trial 3 1156.5 1148.9 1138.9 1115.7 1028.6 908.2 709.8

Trial 4 1156.7 1148.7 1139.3 1115.4 1029.5 908.7 708.8

Trial 5 1156.8 1148.9 1139.5 1115.1 1028.6 908.4 709.0

5
II. Data processing

1. Average frequency (f Ave )

I will calculate the average frequency for each value of ​H​ by adding up the

frequency values that I received for the five trials and dividing by five.

Example calculation for ​H​ = 0:

1156.3+1157.1+1156.5+1156.7+1156.8
5 = 1156.68 Hz ≈ 1156.7 Hz

2. Absolute uncertainty in the average frequency (Δf Ave )

I will calculate the uncertainty in the average frequency for each value of ​H

by subtracting the smallest value of ​f​ from the largest value of ​f​ for the

given value of ​H​ and divide the result by two.

Example calculation for ​H​ = 0:

1157.1−1156.3
2 = 0.4 Hz

Δf Ave
3. Percentage uncertainty in the average frequency (
f Ave × 100 % )

I will calculate the percentage uncertainty in the average frequency for

each value of ​H​ by dividing the absolute uncertainty in the frequency by

the average frequency and multiply the result by 100 %.

Example calculation for ​H​ = 0:

0.4
1156.7 × 100 % = 0.03 %

6
 Table 2. ​Processed data.

H​ (cm) ± 0.2 cm f Ave (Hz) Δf Ave (Hz) Δf Ave

f Ave × 100 %

0 1156.7 0.4 0.03 %

2.0 1148.8 0.4 0.03 %

4.0 1139.3 0.4 0.04 %

6.0 1115.6 0.5 0.04 %

8.0 1029.0 0.8 0.08 %

10.0 908.6 0.4 0.04 %

12.0 709.3 0.5 0.07%

III. Graphing the results

Graph 1.​ Average frequency vs. Height of water level

7
 As explained earlier, the uncertainty in H is ± 0.2 cm. Horizontal error bars are

displayed on the graph to show [Link] the last column of Table 2 we can see

that the percentage uncertainty in the average frequency is negligible, so I

decided not to include vertical error [Link] graph clearly shows that as ​H

increases, ​f​ decreases and that there is a nonlinear relationship between the

variables. This is consistent with the hypothesis and the underlying concepts that

were described in the introduction. Using the Trendline function of the program

that I used to draw the graph, I tried to fit different models to the data set. After

experimenting with various options, the best fit seemed to be a fourth order

polynomial. I will return to this equation in the Conclusion section. The trendline

shown on the diagram (Graph 1) is the graph of this polynomial. The R2 value for

this polynomial trendline is 0.9994. Since this value is very close to 1, it confirms

that the trendline fits the data very well.

Conclusion and evaluation

I. Conclusion

1. Research question, hypothesis and scientific context

The research question was: How does the height of the water level in a

glass affect the frequency of the sound emitted by the glass when the side

of the glass is hit with a spoon? The hypothesis predicted that as the

water level in the glass increases, the frequency of the sound produced by

the vibrating glass will decrease at an increasing, nonlinear rate. The

processed data and the graph clearly show that frequency decreases as

8
 the height of the water level increases. The trendline that seems to best fit

the data is a fourth order polynomial. Based on this, it seems that there is

no obvious, clear mathematical relationship between the variables,

although there might be a more complex connection. Results of similar

experiments show graphs that have the same shape as the graph in this

investigation (Lee 3, Nose 11). Moreover, in one of the experiments, the

relationship between liquid height and frequency is modelled by a fourth

order polynomial. In conclusion, the data and the results of other, similar

experiments confirm that the hypothesis was correct.

II. Evaluation

1. Method

a. Strengths

The data collected in the experiment seems to be precise. First, we

can see in Tables 1 and 2 that during the five trials carried out, the

measured frequency values (except for one: 1030.2 Hz) were all

within the absolute error of the average frequency. High precision is

also confirmed by the negligible percentage uncertainty values in

the measured frequencies (Table 2). In addition, the validity of the

method and the reliability of the data are confirmed by the fact that

both the general tendencies observed in the collected data and the

shape of the graph are very similar to the outcome of experiments

investigating the same topic.

9
 b. Limitations, errors and possible improvements

Source and effects of Significance of error Possible improvements

error

Inconsistency in the Low significance. Create a simple system

strength used when hitting (possibly involving a simple
the glass with the spoon: Considering that the speed pendulum) that can be
this could lead to a and the force we apply used to hit the glass with a
variation in the measured when we rub the rim of the consistent force.
frequency values. glass with our fingers do
not affect the frequency,
we can assume that this is
also true when hitting the
glass with a spoon.

Inaccurate frequency Medium significance. Use two or three different

measurement by the ways to record the emitted
Google Science Journal An upward or downward frequency. These could
App: this software has shift caused in all data include a spectrum
good opinions; however it points would still allow for a analyser, a frequency
is not considered a highly valid conclusion that counter or another App, for
reliable measurement tool. answers the research instance Decibel X.
Based on the data, question. The equation
measurements are precise, connecting the two
but might be lower or variables would be
higher than the actual affected.
frequency values.

Tape measure reading High significance. Use a liquid level sensor to

precision: it was not always measure the height of the
easy to read the water If individual data values are water level in the glass.
level when using a shifted up or down (as
measuring tape, which opposed to the entire
could have led to an graph shifting - see
upward or downward shift previous error), this might
in individual data values. lead to an incorrect
trendline or no trendline at
all.

10
 2. Extensions

There are many ways that this experiment can be taken further. For

example, glasses with different heights, volumes or shapes could be used.

I included two possible, realistic extension ideas in the table below.

Extension 1 research question:

How does the density of the liquid in a glass affect the frequency of the sound

emitted by the glass when the side of the glass is hit with a spoon?

Independent variable Dependent variable Control variables

Density of liquid Frequency of the emitted Type of glass

(g cm − 3 ) sound (Hz)
Height of liquid

Glass cross-sectional
area

Air temperature

Extension 2 research question:

How does the distance between a glass and a measuring device affect the

intensity of the sound detected by the measuring device when the glass is hit

by a spoon?

Independent variable Dependent variable Control variables

Distance between glass Intensity of the detected Type of glass

and measuring device sound (dB)
(cm) Strength with which the
glass is hit

Air temperature

11
 Works Cited

Lee, Junghwan. ​The effect of liquid on the vibrational intensity of a wineglass at steady

state resonance.​ 2018. CheongShim International Academy,

[Link] Accessed 1 April 2020

Nose, Anita. ​Singing Wineglasses.​ 2010. University of Ljubljana,

[Link]

pdf​. Accessed 28 March 2020.

"Pitch." ​Merriam Webster Dictionary​, Merriam Webster Incorporated, 2020,

[Link] Accessed 30 March 2020.

"Singing Glasses." ​Scientific American,​ 2015,

[Link] Accessed 28 March

2020.

12