Introduction to Error Analysis for the Physical

Chemistry Laboratory
January 16, 2012

Introduction

In the physical chemistry laboratory you will make a variety of measurements,

and then manipulate them to arrive at a numerical value for a physical property.
However, without an estimate of the error of these numbers they are largely
useless. Some published numbers in physics and chemistry are accurate to
10 signicant gures or more, while others are only accurate to an order of
magnitude (no signicant gures!). The estimated error of a published number
is a crucial piece of information that must be calculated.
This handout provides a practical introduction to the error analysis required
in a typical physical chemistry lab. Error in an experiment is classied into
two types: random error and systematic error. Random error arises from
unpredictable uctuations in taking a measurement and is the subject of this
handout. Systematic error arises from errors in the equipment or procedure
which cause the number measured to be consistently above or below the real
number. This type of error includes simplications in your model, biased instrumentation, impure reagents, etc. Random error measures the precision
of the experiment, or the reproducibility of a given result. Systematic error
measures the accuracy of a result, or how close a result is to the true value.
Random error can be decreased by increasing the number of measurements you
take, systematic error cannot. The distinction between accuracy and precision
is illustrated schematically in gure 1.

What is uncertainty?

Most chemists have an intuitive idea of what uncertainty is, but it is instructive
to give a more rigorous denition. Suppose you perform an experiment to
determine the boiling point of a liquid, and you measure 32 degrees. How
condent are you in this number? This question could in principle be answered
by repeating the experiment many times and collecting the results. If you
took this large collection of results and counted all of the values that lie within
specied intervals (e.g. between 30 and 30.1, 30.1 and 30.2, etc.) you could
1

Figure 1: Schematic illustration of accuracy and precision. The left-hand target

represents a high precision but low accuracy experiment. The right-hand target
represents a low precision but high accuracy experiment
make a histogram plot (gure 2). You can see that the results are spread over a
range of values. The width of this spread can be quantied using the standard
deviation () of the distribution which is dened as

N
 (fi f)2
=
(1)
N 1
i=1
where fi are the results of your individual experiments, f is the average of your
results, and N is the number of trials performed. The standard deviation gives
limits above and below a measured value in which subsequent experimental
results will probably lie. The number will lie within +/- of the average 68.2%
of the time and within +/- 2 of the average 95.4% of the time.
The spread of a distribution can be used to determine the experimental
uncertainty in your measurement. When a measurement is repeated many times,
the true value is taken to be the average of the values obtained for the many
dierent measurements. The error in the average is then the standard deviation
of the mean which is dened as

f =
N

(2)

where is the standard deviation of the measurement and N is the number of

measurements that were taken. Notice that the standard deviation of the mean
decreases
as the number of points taken increases (though it does so slowly as

1/ N ). Thus, for random error, increasing the number of points taken decreases
the error in the mean.
Most often in the physical chemistry lab you will only perform an experiment one or two times. In this case it is not possible to calculate the standard
deviation of a large number of trials directly. Instead, the theoretical error
can be used. The theoretical error is a measure of the error associated with the
typical use of the glassware and equipment. For example, an instrument such as
2

70
2

60

Counts

50
40
30
20
10
0
27

28

29
30
31
32
33
Boiling point [degrees C]

34

35

Figure 2: Distribution of a series of 1000 boiling point experiment results. The

uncertainty in a single trial is related to the width of the distribution, and is
called the standard deviation of the distribution ().
a volumetric ask will state its uncertainty in its technical specications. If not,
you can make an educated estimate of the uncertainty. If an instrument gives
a digital reading, you can generally take the uncertainty to be half of the last
decimal place. For example, if a digital thermometer reads 25.4 degrees, the uncertainty is 0.05 degrees. For analog instruments, rst read the measurement to
as many signicant gures as there are marks on the gauge, and then estimate
one more signicant gure. The uncertainty should then be estimated based
on how condent you are in the estimated signicant gure. For example, if a
mercury thermometer has marks at every degree, you would read the number of
degrees, and estimate the tenths of a degree. The uncertainty might be plus or
minus 0.2 degrees (e.g. 25.4 0.2 degrees). Typically the standard deviation
of the mean is larger than the theoretical error, however, if both are available
the larger of the two numbers should be assumed.

Error propagation

Often times the measurement of several dierent quantities are used to determine a nal desired value. For example, to determine the number of moles in a
gas sample using the ideal gas law, we would measure the pressure, the volume,
and the temperature. Each of these individual measurements would have both
an experimental and a theoretical error which would need to be considered when
determining our error in the number of moles, n. To determine our error in n
3

we need to propagate the errors in the individual measurements that led to the
nal result. To begin our discussion of error propagation, consider an experiment that measures some quantity x. The result we are looking for is some
function f (x). The measurement of x is subject to some uncertainty bounds,
and the most general case does not assume symmetric uncertainties above and
below x. In this case, the measured value x0 is within the range
x0 < x0 < x0 + +

(3)

where x0 is the measured value of x, + is the uncertainty above x0 , and is

the uncertainty below x0 . The desired property f is then within the range
f (x0 ) < f (x0 ) < f (x0 + + )

(4)

If the uncertainty in x is assumed to be small, the uncertainty in f becomes


df 
f =
x
(5)
dx x0
where x is the uncertainty in x, which we now take to be random, i.e. symmetric. This error can be the standard deviation of the mean or the theoretical error
in x, depending on what is available. If we have a function of many variables,
and if the errors are both small and independent, the uncertainty is

 2
2
f
f
2
f =
x +
2 +
(6)
x x0
y y0 y
We can derive some special cases from equation 6. If a function only contains
addition and subtraction operations, the uncertainty is

f = x2 + y2 +
(7)
If a function only contains multiplication and division operations, the uncertainty is

 2
2
y
x
f = |f |
+
+
(8)
x0
y0
Now that we are equipped with these formulas, we can proceed to propagate
our individual uncertainties. We will illustrate this procedure with an example.
Suppose you want to measure the molar heat of solvation of LiCl in water. This
involves (1) weighing an amount of LiCl, (2) measuring a volume of water, and
(3) measuring the temperature change when the reagent is dissolved. We will
rst calculate the heat of solvation (H) itself, which is expressed in terms of our
three measurements:
C V T
(9)
H =
m/M
Here m is the mass of LiCl, M is the molecular weight, C is the heat capacity
of water per unit volume, V is the volume of water, and T is the temperature
4

change. Note that the function H only contains multiplication/division operations, so we can use the error propagation rule for multiplication and division
(equation 8). The variables we need to consider are m, V , and T . We do not
include M and C in our list of variables, because they are assumed to be known
to much higher (relative) precision than m, V and T . If they were not, we
would have to include them in our error analysis, even if we didnt measure
them. Plugging our variables into equation 8, we have

m 2  V 2  T 2
+
+
(10)
H = |H|
m
V
T
Problem 1 Calculate the heat of solvation of LiCl and its associated uncertainty as discussed above, if the mass is 2.1 0.05 g, the molecular weight is
42.394 0.0005 g/mol, the volume is 0.10 0.02 L, the temperature change is
4.0 0.5 K, and the heat capacity per volume is exactly 4.184 kJ/LK. Dont
include the error in the molecular weight in your calculation.
(Answer: H = 33.79 kJ/mol, H = 8.01 kJ/mol, so you would report the
heat of solvation as 34 8 kJ/mol.)
Problem 2 Perform the same calculation as in the last problem, but this time
include the molecular weight in your list of error propagation variables.
(Answer: You should get the same answer as before, to at least 6 significant
figures! This is why it is often possible to ignore variables known to high precision in your error propagation.)
Sometimes a function may contain both addition/subtraction and multiplication/division, in which case the two rules can be combined. The easiest way
to do this is to break the calculation into steps. Going back to the heat of
solvation experiment, suppose that two separate masses were weighed, and then
both masses were added to the solvent. Now the total mass is m = m1 + m2 ,
and the equation for the heat of solvation becomes
H =

C V T
(m1 + m2 )/M

(11)

The rst step is to calculate the uncertainty in m = m1 + m2 using the error

propagation rule for addition/subtraction, i.e.

2 + 2
(12)
m = m
m2
1
Then simply use the total mass m and its calculated uncertainty, and proceed
as in Problem 1.
Problem 3 Calculate the uncertainty for the previous example if the two weights
were 0.5 0.05 g and 1.1 0.05 g, the volume is again 0.1 0.02 L, and the
5

temperature change is 3 0.5 K.

(Answer: m = 0.07 g, H = 33.26 kJ/mol, and H = 8.78 kJ/mol. Report
as H = 33 kJ/mol, and H = 9 kJ/mol)
Almost all of the error propagation you will do in the physical chemistry lab
will only require the rules for addition/subtraction and multiplication/division.
However, occasionally you might come across a more complicated function, in
which case we need to use equation 6. For example, suppose you have determined G for a reaction and are interested in calculating the equilibrium
constant:


G
K = exp
(13)
RT
Assuming that R and T are known to high precision, we only need to calculate
the partial derivative of K with respect to G:


1
G
K
=
exp
(14)
G
RT
RT
and using equation 6, the uncertainty in K is


G
1
exp
K =
G
RT
RT

(15)

Finally, for the small number of data points typically taken in the lab
(N<20) it turns out that the standard deviation of the mean is not a good
estimate of the error of the mean value. There are several ways to correct the
standard deviation of the mean when only a few points are taken, but the most
straightforward technique is to report a condence interval instead of f. A
condence interval is a range that is likely, within a certain accuracy, to contain
the correct value. For example, if a range is reported to be the 95% condence
interval, there is a 95% chance that the true number will lie within that interval.
The condence limit has the advantage that it is not biased by the number of
data points taken and therefore is a good estimate of the error when the number
of points taken is small. The condence interval is given by

=x
t = x
N

(16)

where x
is the mean value, is the standard deviation, N is the number of
points taken and and t is the t value for N-1 degrees of freedom at the desired
(usually 95%) condence limit. Note that this conversion is done after any error
propigation and is not well dened for values calculated from many measurements where a dierent number of trials are taken for each measurement. In
these cases usually the standard deviation of the mean is recorded. It is up
to you to decide what the correct value for the error should be and to clearly
explain how your error was estimated.

Graphs

In many experiments you will be required to calculate the slope or intercept of

a linear function. For example, the rate of a unimolecular reaction obeys an
exponential rate law
c(t) = A exp(kt)
(17)
where c is the concentration of reagent, k is the rate constant, t is time, and A
is a constant. If you want to calculate k, you will need values for c at dierent
times. Taking the natural logarithm of equation 17 gives
ln c = kt + ln A

(18)

which is a linear equation with the familiar form y = mx + b. In this case, what
interests us is the slope as a function of t, which is our desired rate constant.
Using two data points, (x1 , y1 ) and (x2 , y2 ), the slope and intercept can be
calculated directly:
y2 y1
x2 x1
b = y1 mx1
m=

(19)
(20)

and the uncertainties can be calculated using error propagation. If the error in
x is negligible compared to the error in y, you can calculate the maximum and
minimum values for the slope and intercept as illustrated in gure 3. Drawing
a line through the upper bound of y1 and the lower bound for y2 gives a lower
limit to the slope and an upper limit to the intercept. Similarly, drawing a line
through the lower bound of y1 and the upper bound of y2 gives an upper limit
to the slope and a lower limit for the intercept.
If you have many data points, the average slope and intercept can be
calculated using linear regression. In practice this is always calculated using
a computer program, so the details of the procedure do not concern us. You
simply need to become acquainted with a program capable of performing a linear
regression with error analysis. There is a handout on the chemistry 4581/4591
web site to help you with this. The result of the calculation will give a value
and an uncertainty for the slope and intercept.

Significant Figures

All reported values must be given with the correct number of signicant gures. It is assumed that you have learned the basics of assigning and reporting
signicant gures in your general chemistry lab. However, when reporting uncertainties, you must also consider what numbers you are certain of given the
error you determined from your error analysis. Errors are generally reported
to one signicant gure. For example, if you calculate your error as 0.83, you
are uncertain about the 8 so you are even less certain about the 3. The error
then should be rounded to 0.8, a reasonable measure of how sure you are of
7

60
55
50
45
y

40
35
30
25
20
4

10

11

Figure 3: How to determine the uncertainty in the slope and intercept of two
data points. The bars indicate the uncertainty in the y variable, and the dashed
lines give upper and lower bounds for the line.
your answer. The one exception is if the one signicant gure you would keep
in your error is a one, then you typically would keep two signicant gures. For
example, if your error is calculated to be 0.132, you would report it as 0.13.
This is because the number following the one is a large enough fraction of the
error that it should be reported.
When reporting a number where you have calculated the uncertainty, only
report your number to the same number of digits as you report your error. For
example, if you calculate that the heat capacity of something is 5.86 kJ/molK
but you calculate an error of 0.4 kJ/molK, you should report the value for the
heat capacity as 5.9 +/- 0.4 kJ/molK, even if your signicant gure rules suggest
you should report all three digits. The error in your calculation is large enough
that you cannot be certain of the 6 in this example because you are not even
completely certain of the 8. So by reporting the 6 you would be claiming to
have more information about your heat capacity than you possibly could have
given your level of error.

Further reading
[1] P. Bevinton and D. K. Robinson, Data Reduction and Error Analysis for
the Physical Sciences; McGraw-Hill, 2002.
[2] E. B. Wilson Jr., An Introduction to Scientific Research, Dover, 1990.

[3] J. R. Taylor, An Introduction to Error Analysis: The Study of Uncertainties

in Physical Measurements; University Science Books, 1997.
[4] D. P. Shoemaker, C. W. Garland, and J. W. Nibler, Experiments in Physical
Chemistry, McGraw-Hill, 1996.