A2 Physics Paper 5

Propagation of errors
There is an uncertainty in every measurement that is made, and this
uncertainty can be expressed in an absolute form or as a percentage.
This uncertainty in the raw data leads to uncertainty in all values derived from
that data.
The attempt to quantify the uncertainty in derived quantities is called error
propagation.
1
Revising AS level work on error propagation
At the AS level practical, this treatment of errors was limited to the idea that
the error in a derived quantity z may be found as follows:
if z = xy or z = y/x , then percentage error in z = (percentage error in x) +
(percentage error in y); and
if z = x + y or z = x y, then absolute error in z = (absolute error in x) + (absolute
error in y).
From this, we should conclude that, wherever possible, experimentally-derived
quantities should always have
a numerical value and
an error estimate and
a unit.

2
A2 level work on errors

In evaluating conclusions for experiments at A2 level practical, we need to
find the error (i.e. the uncertainty) in the value of the constants.
To do this, we will need to master these stages:
determine error estimates for calculated quantities using the error estimates in
the measurements
display error estimates in tables of results
display error estimates as error bars on graphs
make error estimates for the gradient and y-intercept of a graph
determine error estimates in the final constant using the error estimates in the
gradient or y-intercept.

3
Displaying error estimates in tables of results

In the AS year, we learnt to estimate experimental errors, but not necessarily
to record them in their tables of results.
At the A2 level we must record absolute error estimates beside every value in
the table of results
This includes both columns of raw data that are measured directly and
columns of derived quantities that have been calculated from the raw data.
It is good practice always to show error estimates beside experimental values
It is also helpful for us to see the error estimates when we are evaluating the
experiment and considering the major sources of error.
From hereon we must include error estimates in all tables of results.
4
Displaying error bars on graphs
Absolute error estimates included in tables of results, can be displayed on the
graphs by drawing error bars.
There are two advantages of drawing error bars on graphs
Firstly, when drawing the line of best fit, the line should pass through (or
very close to) all of the error bars (consider what it means if the line does
not pass through or near the error bar of a data point). If the error bars are
of different lengths at different points, then it is useful to be able to see
this when drawing the line of best fit.
Secondly, the error bars on a graph provide a very visual way of indicating how
precisely the gradient and y-intercept of the graph may be known.

We can plot a horizontal error bar and/or vertical error bar for every data
point if necessary.
5
Determining the error in the gradient and the y-
intercept of a line
In order to determine the error estimates in the gradient and y-intercept of a
line, the usual way that this is done is by drawing, in addition to the line of
best fit, a worst acceptable line through the data points.
The worst acceptable line is acceptable because, like the line of best-fit, it
should pass through or near to all of the error bars.
This means that it would be acceptable as a straight line that passes through
the data points.
However, it is the worst acceptable line because it is the steepest (or
shallowest) of all the possible acceptable lines.
Clearly there are two possible worst acceptable lines the steepest one and
the shallowest one. We only need to draw one and it does not matter which one
we decide to draw.
We then calculate the gradient of both the best-fit line and the worst
acceptable line. The error is the difference between the best and the worst
value.
6
Example
For example, if the gradient of the best fit line is 2.3 and that of the worst
acceptable line is 2.7, then the gradient is equal to 2.3 0.4.
The same method is used to determine the error in the y-intercept of the line.
This means that there will be 2 lines drawn on the graph, and it is important that
these are distinguishable.
The syllabus suggests that the worst acceptable line should either be labeled or
drawn as a broken line.

7
Propagating errors from the gradient or y-intercept
to the final answer
The method used to do this is the same as the method used to propagate errors
from raw data to calculated quantities and to propagate these to their final
answers
8
More on error propagation
At the AS level studies we learnt how to determine the errors in quantities that
are calculated from other quantities by the basic arithmetic functions of
addition, subtraction, multiplication and division.
For more complex mathematical functions, a different approach to error
propagation is required.
The approach is to find the best value and the maximum value, and is
best illustrated by an example.
If x = 7.2 0.6, what is the uncertainty in ln x?
The natural logarithm of the best value of x is ln 7.2 = 1.974.
The natural logarithm of the maximum value of x is ln (7.2 + 0.6) =
2.054.
The difference between these values is 2.054 1.974 = 0.080.
Therefore ln (7.2 0.6) = 1.97 0.08

Two points to note:
Firstly, this approach can be used with any mathematical function, not just with
natural logarithms.
Secondly, it is conventional that the error estimate is normally quoted to one
significant figure, and this should determine the number of significant figures
in the derived quantity.

9
10
Significant Digits/Figures

Significant figures are those digits in a measurement that are known with
certainty plus the first digit that is uncertain.
The degree of accuracy of a measured quantity can only be to the greatest
degree an instrument is able to measure. As such, any digit beyond that
can be measured is irrelevant or insignificant.
When stating any measured quantity, the following general rules should
apply:
a) Experimental uncertainties should almost always be rounded to one
significant figure. Example: g = 9.82 0.03 m/s and not 9.82 0.03385
m/s.
b) The last significant figure should usually be of the same order of
magnitude as the uncertainty. For example x = 6050 30 m/s and not
6051.78 30 m/s.
c) I f the last significant digit is a 1 or 2 in a result, stating the answer
with one extra figure trailing the digit is more appropriate. Example: the
stated answer to calculated value 1415 correct to one significant figure
should be 1400 and not 1000.
More general rules for significant
digits/figures in general
Digits from 1- 9 are always significant.
Zeros between two other significant digits are always significant
One or more additional zeros to the right of both the decimal place and
another significant digit are significant.
Zeros used solely for spacing the decimal point (placeholders) are not
significant.
e.g #OF SI G. DI G. COMMENTS
453 kg 3 All non-zero digits are always significant.
5057 L 4 Zeros between 2 sig. dig. are significant.
5.00 3 Additional zeros to the right of decimal
and a sig. dig. are significant.
0.007 1 Placeholders are not sig.

12
Significant Digits in Addition and Subtraction
When adding or subtracting your answer can only show as many decimal
places as the measurement having the fewest number of decimal places.
Example: When we add 3.76 g + 14.83 g + 2.1 g = 20.69 g
When adding or subtracting measured quantities, the precision of the
answer can only be as great as the least precise term in the sum or
difference. All digits up to this limit of precision are significant.
Example
If x = 3.76, y = 46.855 and z = 0.2. What is the correct value of x + y + z?
Solution
x + y + z = 3.76 + 46.855 + 0.2 = 50.815
largest value = 3.765 + 46.8555 + 0.25 = 50.8705
smallest value = 3.755 + 46.8545 + 0.15 = 50.7595
limit of precision is 0.05 (from x 10
-1
)
= 50.8705 50.7595 = 0.1110/2 = 0.0555 0.05
Since is greater than the limit of precision (0.05) for a one decimal
placing, the third digit ( .8 ) of the answer is itself uncertain.
stating/reporting correct value of x + y + z = 3.76 + 46.855 + 0.2
= 50.815 50.8

Significant digits in Multiplication & Division
General rule:
When multiplying or dividing, your answer may only show as many significant digits
as the multiplied or divided measurement showing the least number of significant
digits.
Example: When multiplying 22.37 cm x 3.10 cm x 85.75 cm we get 5946.50525 cm
3
We look to the original problem and check the number of significant digits in each of the
original measurements:
22.37 shows 4 significant digits, 3.10 shows 3 significant digits,
85.75 shows 4 significant digits.
Our answer can only show 3 significant digits because that is the least number of
significant digits in the original problem.
So 5946.50525 shows 9 significant digits, we must round to the tens place in order
to show only 3 significant digits. Our final answer becomes 5950 cm
3

Example
If l = 31.3, b = 28 and h = 51.85, what is the value of l x b x h?
Solution
l x b x h = 31.3 x 28 x 51.85 = 45,441.34 45,000


What about calculator errors?
Suppose the following readings were taken repeatedly using an ammeter:
2.4 2.5 2.4 2.6 2.5 2.6 2.6 2.5
On a calculator, the mean of the average readings works out to 2.5125
However, as the reading was made to only 2 sig figures, the mean should
also be given to only 2 sig figures i.e. 2.5
As a general guideline, a calculated result should have no more sig fig
than any of the measurements used in the calculation
However, if the result is to be used in further calculations, it is best to
leave any rounding up or down until the end