McGill University

Physics 131
Experiment 0

Uncertainty Propagation Practice

In association with this Exercise, you should read the document “Measurements, Uncertainties, and
Fitting”. Your submitted report should be based on the “Prelab 0 Template”. Both of these
documents are posted in myCourses.

Fall 2023
Introduction - the Situation
You’re sitting in the passenger seat of your friend’s car as they attempt to maintain a constant speed.
You wonder if that is really true, and how accurate your friend’s speedometer really is, and so you
acquire some data and analyse it with those two questions in mind: (a) is the speed constant, and
(b) what is the speed (with an uncertainty, of course)?

Data and Analysis

There are tick marks on the road every 100 m, which you use to measure the distance you travel. You
use the stopwatch on your phone to measure time. You estimate the uncertainty in the distances as
approximately one car length, about 5 m. Your starting estimate for the uncertainty in time is the
human reaction time, 0.3 s, but it decreases with subsequent measurements because you get better
at anticipating the roadside markers.

You will download your data for this exercise from myCourses.

You can create a table similar to Table 1 with your data (do not use the data in Table 1; it is for
illustration purposes only!).

Measurement t (s) δt (s) d (m) δd (m)

a 0.0 0.3 0. 5
b 10.1 0.2 100. 5
c 20.5 0.2 200. 5
d 30.1 0.1 300. 5

Table 1: Total time t and total distance d the car traveled. This data is illustrative only (your data
will come from myCourses).

There are two obvious approaches (can you think of others?) to this analysis and the questions we’ve
set out to answer. The first would be to fit the data, which is a set of x-vs-t points, to a linear
function: x = (slope)× t + offset; the slope parameter will be the speed. The second approach would
be to recast the data as a series of speeds during the multiple intervals (a to b, b to c, etc), then
check if those speeds are consistent (within their uncertainties) with being constant.

You will perform both approaches, and report on your results, in this exercise.

Method I: Plot and fit x-vs.-t data

1. Put your data into a table with the same structure as Table 1. Make sure this table is in
your lab report.

2. Create a graph of your data, using the curve.fit website. Set the fit type to “Linear” and
choose “Table” for the uncertainty (δx and δy) inputs. Enter your total time values and
uncertainties as x and δx, and your total distance values and uncertainties as y and δy. Enter
a graph title and appropriate x- and y-axis labels, and press “Fit” to perform the fit and see
the results. Take a screen shot or download the results and put this in your report

1
 (including the plot of the fit, the χ2reduced value, and the parameter estimates with
uncertainties.
3. Comment on the quality of the fit (you will want to refer back to the “Measurements, Uncer-
tainties, and Fitting” document). In particular, look at and comment on the residuals, and the
value of the χ2reduced quantity.
4. If the fit is good, extract the resulting speed measurement, with uncertainty, and clearly note
it in your report.
5. Answer the two questions we set ourselves at the top of this page.
6. Return to curve.fit and perform the same fit, but setting b = 0 (thereby forcing your fit to
go through the origin). Note the new χ2reduced value and the new value of the slope in
your report, and comment on the comparison with the original values. This is a case
where you don’t really need to include a figure of the new fit; the reader is convinced by this
point that you know what you’re doing with curve.fit and so simply quoting these new values
is sufficient. *** Note: when you are performing Labs 1-4 and plotting your data, do not set
b = 0.

Method II: Calculate and average interval data

For the second method, you will need to calculate the speed data from each interval of your data (a
to b, b to c, etc.). Create a table (make sure to include it in your report!) similar to Table 2
below.

To calculate the uncertainties for your Table 2, a summary of useful equations is given at the end
of this document; you might also need to refer to the “Measurements, Uncertainties, and Fitting”
document. The distance Δd and time Δt for a particular interval are the result of subtractions so
the uncertainties will add in quadrature. You will also need to calculate the resulting uncertainty in
the speed value Δd
Δt .

Include at least one set of sample calculations in your report (i.e, the detailed calcula-
tions for one interval).

Interval Δt (s) δ(Δt) (s) Δd (m) δ(Δd) (m) v (m/s) δv (m/s)

1 10.1 ? 100. ? 9.9 ?
2 10.4 ? 100. ? 9.6 ?
3 9.6 ? 100. ? 10.4 ?

Table 2: Change in time Δt, change in distance Δd, and average speed v of the car for each mea-
surement interval. Again, this is example data: you will need to calculate this data from the raw
data give to you through myCourses.

Your task is now to decide whether these values of v are consistent with being the same, and if so,
what the mean value is. You could use the math of a weighted mean to calculate the average value
- but notice that approach would not give you any statement about whether the individual speed
measurements were consistent with one another. A different approach is to let curve.fit do the
work:

2
 1. Input your values of v and δv as y values, with x values of 1., 2., 3., 4., etc. and no x uncertainty.

2. Choose a linear fit, and fix the m parameter to zero (ie, forcing the fit to have no slope - that
is, to fit to a constant value).

3. Paste the resulting fit - showing clearly the plot, the residuals, and the χ2reduced
value - into your report and comment on the quality of the fit (as above, making reference
to the residuals and the χ2reduced value).

4. If the fit is good, extract the resulting speed measurement, with uncertainty, and clearly note
it in your report.

Based on your observations in [3.] and [4.], answer (again) the two questions that we set ourselves
at the start of this document.

Conclusion
You’ve now analysed this data by performing three fits: a fit to a linear function x = (slope)×t+b,
a fit to a linear function with no offset x = (slope)t, and a fit to a constant velocity of the interval
data.

Given the results of these analyses, give your best answers to the two questions asked in the Intro-
duction; make sure this is included in your report.

Information to include in your report

A template report specifically for this pre-lab exercise is posted on myCourses. Please use it! At a
minimum, your report should have:

1. A table containing your data from myCourses;

2. A screen shot of the curve.fit plot from the first method;

3. A brief discussion of the quality of the fit in the first method, and your conclusions from the
two fits;

4. The equivalent of Table 2 above, with derived interval data;

5. At least one set of detailed sample calculations for the uncertainties in Table 2;

6. A screen shot of the curve.fit plot from the second method;

7. A brief discussion of the quality of the fit in the second method, and your conclusions from
that fit;

8. Enough narrative text throughout so that you can follow what you did when you look back at
this report in six months;

9. Your best answers, for each of the two methods, to the two questions asked in the Introduction.

3
Uncertainty Equations:
Rule Mathematical Relation Uncertainty
q
Addition/Subtraction z=x+y δz = δ2x + δ2y
Multiplication with a constant A z = Ax δz = |A|δ
sx 
δx 2
 2
δz δy
Multiplication (2 independent variables) z = xy = +
|z| s x y
 2  2
x δz δx δy
Division z= = +
y |z| x y
δz δx
Power (including square roots) z = xn = |n|
|z| |x|
P
zw 1 1
Weighted Average zavg = Pi i i , wi = 2 δavg = pP
i wi δi i wi