Mathematics

Uncertainties

What is an uncertainty?
Physics is a quantitative science, and experimental work is based on measurements. There is never a perfect
or absolute measurement, and a degree of uncertainty is associated with all measurements. Uncertainties
are the result of human involvement in making measurements, the instruments or equipment being used
or environmental factors that can be difficult to control during data collection. Care should be taken to
recognize and reduce uncertainties. It is important to record all measurements with a corresponding
uncertainty.

Human involvement
Subtle differences can be introduced when humans take the same measurement several times. Reaction
times for starting and stopping the timer are not always the same nor are they symmetrical, for example.
Repeating the timing for the same distance may reveal an uncertainty larger than the basic precision of the
measurements. Uncertainties due to reaction time often become apparent through repeated
measurements under the same conditions.
Parallax error is caused by reading an analogue device at an angle to the scale markings. Depending on
how observations are made, a degree of uncertainty caused by parallax can be introduced. If a
measurement on an analogue scale is always taken by looking from the left of the scale, then
measurements may be slightly larger than they should be. This can be reduced through good experimental
practice and ensuring devices are read in an optimum way. Parallax error may appear as a systematic shift
on a graph.

Instruments and equipment

Instruments used to make measurements will have an inherent uncertainty and this is often stated by the
manufacturer. Repeated measurements of the same system may show some fluctuations in the data due to
the instrument. In addition, measurements of a given quantity using two (or more) devices may show some
differences. Any one measurement is not necessarily wrong, but all devices have a degree of uncertainty
associated with them.

Environmental factors
Environmental factors could include temperature, air resistance or background radiation. For example, in an
experiment to measure how the coefficient of restitution of a squash ball depends on the temperature of
the ball, the time lapse between heating the ball and making the measurements will cause a change in the
temperature of the ball. While such sources of uncertainty cannot be entirely eliminated, steps can be taken
to reduce their impact.

Uncertainties in raw data

Determining the sizes of uncertainties
During the Diploma Programme (DP) physics course, students will develop skills on how to estimate
uncertainty. The IB does not prescribe one method and the course allows for a range of approaches to be
applied. It is important that when determining the size of any uncertainty, a student should be able to
justify their choice based on the measuring device and how well they were able to make the measurement.
Experimental example

Physics teacher support material 65

Uncertainties

As part of an experiment to determine the acceleration of freefall, a student measures the time it takes for a
ball to roll a given distance down an incline as 2.3 s. In determining the sizes of the uncertainties, it should
first be recognized that there is a limit to the precision of the stopwatch and a limit to the precision of the
metre rule.
All measurements involve two reference points. The first reference point for a time measurement would be
at 0. The time read-out of 0.0 s could require a brief interval of 0.067 s before it begins timing. That is, the
zero reading is really –0.067 s. Or, when the timer starts it could already be at a time of 0.0123 s. It can be
estimated that any zero reading has an uncertainty of one-half the least count.
The second reference point is the measurement of the event. A measurement of 2.3 s could have been
2.2876 s rounded up or 2.3210 s rounded down. In either case, the measurement could be off by up to one-
half the least count, or ± 0.05 s. However, this uncertainty would be inconsistent with the one-decimal place
precision of the measured time.
In the worst possible case, twice one-half the least count equals plus or minus the least count. The timing in
this example would be (2.3 ± 0.1) s, as noted in the table on measurement and least count. Uncertainties
associated with a single measurement must be expressed to the same degree of precision as the
measurement. This least count of the timer is the minimum uncertainty that should be recorded for a single
measurement. This is the minimum uncertainty, although often it is greater and this normally becomes
evident with repeated measurements.

Measurement and least count

Measurement Least count

2.3 s 0.1 s
2.34 s 0.01 s
2.345 s 0.001 s

The same rule applies to analogue devices such as a metre rule.

Significant figures in raw data must be recorded precisely and consistently, as seen in the table on correct
and incorrect expressions of a length and its uncertainty.

Correct and incorrect expressions of a length and its uncertainty

Correct Incorrect

(87.4 ± 0.2) cm (87.4 ± 0.05) cm

(87.4 ± 2) cm

Propagation of uncertainties
Processing data may involve several mathematical operations, such as sum and difference, product and
quotient, exponential powers, and trigonometric and logarithmic functions. The required basic rules are
provided in the Physics data booklet.
All processed data should be presented clearly. Basic mathematical operations do not need to be
illustrated. More specific and complicated calculations should be illustrated with a sample calculation.

Uncertainties and repeated measurements

The quality of data is revealed by the scatter of data points above and below the best-fit line. These are
called “random uncertainties” and they demonstrate that measurements never have zero uncertainty.
Repeated measurements are the norm in physics, but repeated measurements do not reduce random
errors in the measuring process. Repeated measurements reduce the standard deviation of the mean, but
statistical analysis is not expected in DP physics.

66 Physics teacher support material

 Uncertainties

When making repeated measurements, the uncertainty of the mean can be determined by taking one-half
the range between the maximum and minimum values. Symmetry can be assumed, so that the plus and
minus values are the same.
For example, the extended length of a spring under the same tension is recorded five times, as displayed in
the table on the length and mean length of a spring.

Data on length and mean length of a string

Length of spring, L (cm) Mean length of spring,

ΔL = ± 0.1 cm L / cm

11.5 11.16
11.0
11.1
11.0
11.2

Range = maximum − minimum

= (11.5 − 11.0) cm
= 0.5 cm
One-half the range is 0.25 cm.
Raw data values, the mean value and the associated uncertainty should be expressed to consistent degrees
of precision by recording the value to a consistent number of decimal places. As all raw data are recorded to
one decimal place, a second decimal place in the mean and uncertainty is not justified. Thus, the mean and
uncertainty are best expressed as (11.2 ± 0.3) cm.
However, if this value is used to plot a graph with uncertainty bars, then the additional decimal place can be
retained to avoid compounding rounding errors during data processing. Rounding in preliminary
processing steps can distort the resulting uncertainty. It is common practice in physics to record one or
more additional decimal place during processing, keep additional decimal places and only round at the
end.
If repeated measurements reveal no differences, then the least count should be taken as the uncertainty. If
one-half the range reveals an uncertainty that is smaller than the individual raw data uncertainty, then the
uncertainty in the mean must be taken as the larger of the two values.
Processing uncertainties with trigonometric functions, exponentials, logarithms and other functions can
result in asymmetric uncertainties. In these cases, the uncertainty bar above and below the datum point
would not be equal. Each extreme would be calculated [e.g. log (x + Δx) and then log (x – Δx)]. However, for
the purposes of internal assessment, this subtle detail is not expected and symmetric uncertainty bars are
acceptable.

Significant figures in uncertainties

Most uncertainties should be expressed as a single digit and to the same decimal place as that of the
quantity value. However, if the uncertainty begins with 1, then two significant figures are acceptable, as
seen in the table on correct and incorrect significant figures in uncertainties.

Correct and incorrect significant figures in uncertainties

Correct Incorrect

(12.2 ± 0.4) m (12.2 ± 0.36) cm

(14.23 ± 0.12) s (14.23 ± 0.1) s

Physics teacher support material 67

Uncertainties

Processed data tables

Quantitative data should be tabulated. It is good practice to include a heading with the name of the
quantity (e.g. distance), the symbol of the quantity (d), the unit of the quantity (cm) and the uncertainty for
this unit (Δd = ± 0.2 cm).
Clear and precise tabulated data can include a combination of processed and raw data to help focus the
processing. Explanations of the uncertainty and processing should be provided in the text.
Experimental example
A student measures the period of a simple pendulum of a given length and records these observations in a
table.
Controlled variables can be included outside the table, as seen for the table showing an example of
properly presented raw and processed data.

An example of properly presented raw and processed data

Pendulum length, L (cm) = 80.0 ± 0.1
Time for Period T /s Mean period T /s Mean period squared

10 periods T10 /s ΔT = ± 0.002 s ΔT = ± 0.0425 s T 2 /s

ΔT2 = ± 0.1498 s2
ΔT10 = ± 0.01 s

17.90 1.790 1.7618 3.1039

17.38 1.738
18.00 1.800
17.66 1.766
17.15 1.715

Range maximum − minimum

=
2 2
(1.800 − 1.715)
= s
2
= 0.0425 s
Mean period T = 1.7618 ± 0.0425 s
= 1.7618 ± 2.4123%
To determine uncertainty in T2, the percentage uncertainty in T is doubled.

T = 1.7618 ± 2.4123%
T 2 = (1.7618)2 ± 4.8246%
= 3.1039 s2 ± 4.8246%
= 3.1039 s2 ± 0.14975 s
= (3.10 ± 0.15) s2
Rounding to appropriate significant figures and precision takes place only at the end of the processing.

Data tables using spreadsheets

A spreadsheet can be used to present and process data, and a screenshot may be included in submitted
work. It is good practice to note that a spreadsheet has been used. When using a spreadsheet, it is not
always possible to include uncertainties or other details of the measured quantities in the headings. In
presenting the data, any missing information can be mentioned in the text close to a screenshot of the
spreadsheet data. See, for example, the figure on data processing using a spreadsheet.

68 Physics teacher support material

 Uncertainties

An example showing data processing using a spreadsheet

Student comments could include the following.

The length uncertainty was ± 0.1 cm and the minimum uncertainty for 10 periods was ± 0.01 s. Using the
spreadsheet, the time for 10 periods was divided by 10 to obtain the time of a single period. Then, the mean
value of the five measurements was determined to be 1.7618 s. The spreadsheet was used to determine half
the range of the individual period measurements.
The “Show Formula” function in Microsoft Excel can also be used to illustrate data processing, as displayed
in the figure on the screenshot of this function.

Screenshot of the Show Formula function turned on during data processing using a spreadsheet program

Graphing uncertainties
Uncertainty in raw data should be propagated and can be shown as uncertainty bars on graphs.
Uncertainties in processed data should be considered in the interpretation of the data. When a linear graph
is established and the gradient is meaningful, uncertainty bars should be used to establish the uncertainty
in the value of the gradient.
Automatic gradient determination
Software that automatically determines gradients may be used, but the student must make clear that they
understand the process taking place.
Experimental example
A student conducted an experiment using video to analyse the projectile motion of a ball launched
horizontally from a table. The student clicked on the image position for consecutive time intervals. The
software determined the speed at each position, as shown in the table on time, X and Y positions and
vertical speed. A graph on vertical speed against time was generated automatically, and the gradient was
determined by the software. Acceleration due to gravity is the vertical acceleration of the projectile. The
student described what was done. An equation or sample calculation is not required for the experimentally
determined value of acceleration, but the student must explain how the gradient represents acceleration.
Uncertainties should still be addressed and stated in a format consistent with the final value.
The student could use this result as part of an investigation into a factor that affects the acceleration of a
falling object.

Physics teacher support material 69

Uncertainties

Determination by software of the speed of the ball at each position

Time (s) X position m Y position m Vertical speed /m s−1

0.000 −0.001 1.081 0.000

0.067 0.101 1.091 0.149
0.133 0.200 1.084 −0.106
0.200 0.297 1.084 0.000
0.267 0.396 1.059 −0.373
0.334 0.495 0.993 −0.992
0.400 0.594 0.880 −1.694
0.467 0.693 0.728 −2.279
0.534 0.793 0.526 −3.028
0.600 0.893 0.284 −3.628
0.667 0.986 0.000 −4.258

Graph of vertical speed against time

The slope of the line is calculated as follows.

∆y ∆v
m= = = a
∆x ∆t
= − 9.7804m s−2
There is no need to add the R2 value to graphs.

Uncertainty bars
Uncertainty bars are added to the dependent variable on graphs. This can be achieved using graphing
software. The size of the uncertainty bar should reflect the size of the uncertainty for that data point and be

70 Physics teacher support material

 Uncertainties

based on the spread of data obtained from repeated measurements. It is acceptable to assume that
uncertainty bars will be symmetrical around each data point.

Data outliers
When processing or graphing experimental data, there may be a data point that does not fit the overall
pattern or trend of the other data, as seen in the figure on an example of an outlier on a graph. This is a
potential outlier. In physics, outliers are most likely to occur as a result of human error such as a
methodological fault or an irregularity in the equipment or environment. These can be recognized and
corrected at the time of collecting data. The quantity in question can be remeasured. When adding the
best-fit line or best-fit curve, a decision needs to be made whether to include or exclude any possible
outliers, and a justification should be provided.

An example of an outlier on a graph

The graph illustrates an example of a genuine outlier. The student reasoned that a linear relationship exists
between the variables, and that it did not make sense that one point should be so far off the obvious trend.
The approach used in DP mathematics to identify outliers is not required in DP physics.

Best-fit line or best-fit curve

A best-fit line or best-fit curve should be added to all graphs, and the choice should be driven by the
relevant physics theory. The best-fit line or best-fit curve does not need to pass through all the data points,
but it should pass through—or be close to—all uncertainty bars. Otherwise, the line would not be a best-fit
line, as shown in the figure on the inappropriate line added to a graph.

Physics teacher support material 71

Uncertainties

Inappropriate line added to a graph

A polynomial curve that passes very close to all the data points can be added, as seen in the figure on data
points joined by a polynomial curve. However, the nature of the curve is unlikely to be supported by the
laws of physics.

Example of all data points joined by a polynomial curve

The equation of the curve provided by Excel is y = 0.0253x5 − 0.4229x4 + 1.3446x3 + 11.207x2 − 82.835x +
170.75.
The final example shows a curve that passes close to each data point and through the uncertainty bars.

72 Physics teacher support material

 Uncertainties

Curve passing through each data point and the uncertainty bars

Straight line graphs

In some cases, plotting the experimental data will generate a linear fit. When this occurs, the gradient and
intercept can be determined and related to the relevant theory in the conclusion.
A non-zero intercept may suggest a systematic error that should be discussed in the evaluation, or it may
relate to a physical quantity (see in "Examples of uncertainties in processed data", example 2).
In cases where a best-fit curve has been drawn, the student must decide if the data can be linearized easily.
Relevant background theory, for example squaring reciprocals of one variable, or dimensional analysis, can
provide direction, although this normally requires ideal data. It should be noted that the linearization of
data can be a useful exercise but is time consuming and may not be worth pursuing at the expense of other
tasks. For instance, consider the data from the above example. The reciprocal of one of the variables is
taken and these data are then plotted. The uncertainties are processed and graphed, as seen in the figure
on linearizing data by taking the reciprocal of one variable.

Graph obtained after attempting to linearize data by taking the reciprocal of one variable

An alternative approach also exists, as seen in the figure on plotting the log (dependent variable) versus the
independent variable.

Physics teacher support material 73

Uncertainties

Alternative attempt to linearize data by plotting the log of one variable against the other variable

Maximum and minimum lines of fit

If a straight line has been drawn on a graph, uncertainty bars should be used to draw two further lines: one
showing the maximum permissible gradient, and the other showing the minimum permissible gradient. In
doing so, all data points and corresponding uncertainty bars should be considered, not simply the first and
last points. This allows for the determination of the gradient and intercept with uncertainty.

Examples of uncertainties in processed data

Comparison with literature values (example 1)

Experiment Indicators
A student investigated the ohmic behaviour of an The accepted value of the resistivity of aluminium is
electrical conducting material. The results were used 2.8 × 10–8 W m. A textbook reference or online
to determine the resistivity. source must be given for this value.
The experimental value of the resistivity of
The experimental range of the experimental value
aluminium was calculated to be (3.0 ± 0.5) × 10–8 W
includes the accepted value.
m.
The percentage difference is calculated as follows.
2.8 − 3.0
= × 100%
2.8
0.2
= × 100%
2.8
≈ 7%

Learning opportunities

Calculate the percentage uncertainty in the experimental result and comment on this in the evaluation.
0.5
= × 100%
3.0
≈ 17%

74 Physics teacher support material

 Uncertainties

Comparison with literature values (example 2)

Experiment Indicators

A student investigated the internal resistance of a 6- The y-axis intercept, corresponding to the negative
V battery. They measured the voltage and current of internal resistance, is –(0.5925 ± 0.0866) Ω. That is,
with digital meters for a variety of resistance loads. r ± Δr = (0.59 ± 0.09) Ω
Following relevant theory,
Note: For clarity, uncertainty bars and max–min lines
ε = I(R = r) are not shown on the graph.
ε
=R+r The student was able to find the manufacturer’s
I
ε specification sheet online and discovered that the
R+r= internal resistance of a new battery would be
I
ε 0.483 Ω (no uncertainties were listed). Background
R= − r
I information states that temperature affects the
1 internal resistance and that the internal resistance
R=ε − r
I increases as the battery is used.
y = mx + c
The student speculated that the battery’s internal
The student constructed a graph of the calculated resistance may have increased over repeated
load resistance R against the reciprocal of the measurements. The experimental value of 0.59 ±
1 0.09 Ω compared to 0.483 Ω seems reasonable,
current . This graph showed only the first four data
I given this information.
points so that the view of the y-intercept was
enhanced.

Theory indicates that the y-intercept represents the

negative of internal resistance.

Uncertainties were determined and propagated

correctly but were not visible on the graph.

Graph

Learning opportunities
The experimental value of internal resistance and its uncertainty range did not include the manufacturer’s
value (rmin = 0.50 Ω > rnew = 0.48 Ω). The student’s results were close and of the same order of magnitude,

Physics teacher support material 75

Uncertainties

and the difference was addressed in a plausible way. However, this aspect should have been noted by the
student.

Using theory to determine analysis

Experiment Indicators
A student conducted a simple sonometer laboratory The student sketched the results in a graph of
exercise and quoted the known equations relating tension against fλ, which refers to wave speed.
frequency f, tension T and mass per unit length μ. In They quoted an R2 correlation value of 0.985 and
the background and theory section, the equation T thereby claimed that their hypothesis of a “direct
= v2 μ was stated. relationship between tension and frequency” was
99% certain.

Graph

Learning opportunities
While expressing speed in units “Hz cm” is acceptable, there is a contradiction between the known theory,
the graphical analysis and the conclusion shown here. If the student had understood the known theory,
they would not have forced a linear fit. The student should, therefore, make a comparison to the known
theory.
No attention was paid to the y-intercept other than calculating the gradient range. No attempt was made
to explain what negative tension means (not even as a systematic error) and how this would cause a
resonating frequency from 0 to about 75 Hz. An explanation of the physical meaning of the results was
also missing.

Large uncertainties: potential problems

Experiment Indicators
A student investigated the relationship of projectile A graph of launch angle against distance was
range and launch angle. plotted with uncertainty bars.
They obtained a range of angles and propagated A single best-fit line was applied to the whole data
uncertainties, as seen in the uncertainty bars. set, and the student stated that as the launch angle
increases, the distance increases.

76 Physics teacher support material

 Uncertainties

The student’s hypothesis was qualitative and only

stated that the range would increase with the angle.
In the background research, the student mentioned
the textbook theory of maximum range at 45° if
there is no noticeable air resistance.

Graph

Learning opportunities

At ± 2 m, the uncertainties appear large. While a linear fit lies within the uncertainty bars, this is not
justified by theory. Touching all the uncertainty bars is not a justification here, as the data scatter is not
linear. If the student had constructed the minimum and maximum gradients, the values could have
ranged from positive to negative, including a zero gradient.

The student should have commented about the y-intercept at 7 m. How does this relate to the
experiment?

The student should have referred to the theory while conducting the analysis. One thing is obvious from
the graph: the maximum range was obtained at about 45°.

Further data processing could have been carried out by referring to the textbook equation for projectile
range and drawing a graph of range against sin(2θ). The ideal result would be a line with the maximum
range at sin(2θ) = 1, which is 45°.

Physics teacher support material 77

Uncertainties

The ideal graph: Range vs sine of twice the angle in degrees

Uncertainties in scientific investigation

Addressing the conclusion criteria
This section concerns addressing uncertainties in the context of the criteria for the conclusion of the
scientific investigation.

Precision and accuracy

Video analysis of projectile motion

A student analysed projectile motion using video analysis, as shown in the figure on horizontal and vertical
positions versus time. The results were satisfactory, but the methodology contained some limitations or
assumptions.
Students should be familiar with resolution, frame rate or timing, scale and parallax. Students must address
any issues pertaining to their research questions, aspects that become apparent once they have analysed
the data.

78 Physics teacher support material

 Uncertainties

Analysis of horizontal and vertical projectile motion using video

Possible sources of uncertainty include the following.

• The camera had a fixed position, and there were parallax issues (leading to a lack of accuracy) when
the ball’s path moved from the left to the right of the camera. Increasing the distance away from the
projectile path would reduce the parallax, but this would make the object smaller and make it more
difficult to locate a fixed point.
• The projectile path was assumed to be in a single plane. In real life, this is not the case, and this
resulted in a lack of accuracy. A second and third camera could record the other two dimensions and
determine if the motion out of a single plane was significant.
• The images revealed fuzzy edges of the ball. The limit of resolution affected the precision of the
measurements. An increased frame rate would improve the precision.
Due to the random error exhibited on the graph, it was clear that the last issue noted here would be the
most significant.

Pendulum: A systematic shift

In a simple pendulum investigation, the student graphed the period against the square root of the length
and obtained a straight line. However, the line did not go through the origin as expected, as shown in this
figure.

Physics teacher support material 79

Uncertainties

Graph depicting systematic shift

The student should appreciate the size and direction of the systematic shift in the data and consider this
aspect in the evaluation section.
The student should consider either of the following.
• The y-intercept is 0.1201 s2, or there is a systematic shift in the time of 0.346 s, which suggests that all
measured times are approximately 0.35 s shorter than they should be.
• The x-intercept is −0.0325 m (found by solving for x when y = 0), which suggests that all measured
lengths are approximately 3.3 cm shorter than they should be.

Addressing the evaluation criterion

This section concerns addressing uncertainties in the context of the evaluation criterion of the scientific
investigation.
As part of the evaluation, the student must refer to issues identified as weaknesses or limitations. These are
issues that can apply to the quality of the data or the specific details of the student’s research question and
methodology. Based on these weaknesses or limitations, the student should suggest improvements.
Generalities such as “take more measurements” or “improve precision” should be avoided as the context is
missing. Issues about the quality of data and sources of uncertainties should be addressed. Any suggested
improvements must be plausible and specific.
The following is a detailed example where the student explains the impact of methodological weaknesses
and offers realistic improvements relevant to these weaknesses.

Refractive index
A student investigated how the temperature of water affects the refractive index of light. A standard
textbook method was followed. A fish tank with a water heater was set up, and a laser was secured at a 45°
angle above the water surface. A ruler was fixed at the bottom of the tank (under water), and the distance
from where the laser beam hit the ruler was read off.
The water depth was measured and, using the two perpendicular lengths, the angle of refraction was
determined. The 45° incident angle remained fixed for all temperatures. This method was repeated for
every 5°C for a temperature range from 5°C to 90°C, as seen in the table on temperature versus
experimental value of refractive index.
The measured distances were a few centimetres, with a claimed precision of 1 mm. The incident angle was
measured with a protractor with an uncertainty of 1°, and the refracted angle was calculated.

80 Physics teacher support material

 Uncertainties

The student discovered that the refractive index of water decreases as temperature increases, as displayed
in the table on temperature versus experimental value of refractive index. The results were feasible and
suggested a curve that resembled known theory.

Temperature vs experimental value of refractive index for water (only selected data are shown)

Temperature / °C Experimental value of refractive index ± 0.02

5 1.238

20 1.287

90 1.192

Upon closer examination, however, the student began to question their methodology and the results.
The textbook value of the refractive index of water at 20°C is 1.333. The accepted value was not included in
the experimental range. The student believed an error must have occurred in the either method or the data
analysis.
The student noted that other factors cause a variation in the refractive index, including:
• the wavelength of light
• effect of atmospheric pressure on the refractive index
• purity of water.
For all these factors, only the fourth and fifth decimal places would be affected.
The student calculated the range of their refractive index values. The results showed a change in the index
of n90° − n5° = (1.192 − 1.238) = −0.046. In contrast, the theoretical change for the same temperature range
should have been only −0.012. The student’s range of refractive index was nearly four times greater than
the theoretical range. They realized that the problem was not a systematic error.
The following is a list of weaknesses in the method and relevant improvements.

Methodological weaknesses Improvements

The uncertainty in the distance was Aligning the zero end of the ruler while under water required the
stated as 1 mm. zero point to be directly below the interface point on the surface of
the water.
The laser beam itself was a few millimetres wide.
Looking through the water revealed some distortion of the image.
These issues suggest that a larger uncertainty should be considered.
Precision is low when calculating Vary the incident angle and, with the corresponding refracted angle,
the refractive index from a single use a graph to determine the best-fit line that represents the
set of angles. refractive index value. This would improve precision.
The use of a fish tank and the low Replace the fish tank with a hollow transparent prism on a turntable.
precision of the angle Measure the laser beam deviation several metres across the room.
measurements are also weaknesses. The uncertainty would be a significantly smaller percentage of the
actual measurement.
A polynomial best-fit curve of Relate the refractive index to the water density and look for a linear
refractive index against function.
temperature is only qualitatively With an improved method and precision, a more accurate result
related to known theory. might be obtained.

Physics teacher support material 81

Uncertainties

Methodological weaknesses Improvements

Graphing theoretical and experimental values on the same graph will

highlight any issues.
Heating requires agitation of water Start with hot water and let it cool slowly to room temperature. Then,
to equalize temperature start with cold water and let it warm up to room temperature. Do not
throughout the tank. This affects agitate the water and keep the thermometer in a fixed position.
the surface of the water and the
incident laser beam.
A non-linear systematic error could A revised method (with a prism and turntable) should be used to
occur in the index calculations. measure multiple data sets to find the index at a given temperature
Some unidentified quantity affects and eliminate the compound error.
all the calculations but in an
increasing amount.
The fish tank method did not
consider the slight change in water
depth, which was assumed not to
change during the experiment. This
wrong assumption could be the
source of the non-linear error.

82 Physics teacher support material