Copyright © Michael Richmond.
This work is licensed under a Creative Commons License.
Phys 273 -- Notes on Lab 1
After reading lab reports submitted for this lab
in past years,
I've noticed a set of common omissions.
Most of them are in the Error Analysis section.
Here are some hints for this -- and all! -- experiments.
- After you have identified a source of error,
ask the question: "How big is this error?"
You can go into several
layers of detail:
- Which sources of error are the largest?
At the very least, name the biggest.
- Quantify the amount of error from each
source. If the measurement of time was good to 1 second,
and the total time was 2100 seconds, then the percentage
uncertainty in the time measurement is (1/2100) = 0.05 percent.
That's very small!
- Make a list of the various sources of error, in order from
most severe to least.
- Note that some measurements aren't limited by the equipment, but
by other factors. For example, if one person closes a switch
and other starts a stopwatch, it's possible that the two people
don't act at exactly the same time. Even though the stopwatch
may read time to 0.01 seconds, the true uncertainty in the
measure time may be 0.5 second, or 1 second, due to some
mistake in synchronizing the two actions. If you have to
guess at the size of some error like this -- go ahead
and guess! Just write explicitly, "I'm estimating the uncertainty
in the timing process to be 1 second, for the following reason: ..."
If you can come up with some way of measuring such errors,
even better.
- Always ask, "What's the effect of this error on the final result?"
If you think that you may have cleaned some of the deposited
copper off the plate by mistake, then you should follow this
to its consequence: if the measured change in mass is too small,
then the calculated value for e would be too big
(because you have to divide by "change in mass" to get e).
Say so.
If you can determine the size of the error in the final result
due to an error in the procedure, even better. For example,
you might decide that the current was 5 percent larger than
you mistakenly measured. This would make the final value for
"charge on electron" 5 percent too big.
- If you identify an error, try to figure out a way to
eliminate it, or at least minimize it, for future experiments.
If the ammeter is the biggest source of error, perhaps you need
a better ammeter. Say so.
- If you are able to propagate uncertainties into the final result,
check your estimated uncertainty against the difference between
your final result and the expected value.
For example, if you propagate errors from the stopwatch,
ammeter, and balance, and find that the final uncertainty
in "e" was 4 percent ... but also determine that your answer
differs from the expected value by 10 percent ... then
something fishy is going on! Either you underestimated the
uncertainty in your measurements, or there was some source
of error you didn't consider.
You should at least compare these two checks on your answer,
and comment on their agreement or disagreement.
It would be even better to go back and try to figure out
extra sources of error if you need to do so.
Copyright © Michael Richmond.
This work is licensed under a Creative Commons License.