We used the "real thing" where I taught, oil, etc., but we didn't load the
apparatus with a radioactive source for changing the charge on an
individual drop. We went through the whole calculation using Stokes' Law,
too.
This was one of a "circus" of experiments in modern physics because we
didn't have class sets of any of it, so every pair of students was doing a
different experiment over several weeks, often with two or three
experiments per week.
First of all, I had students do a "black box" experiment of determining
what could be said about the objects in a whole set of about 20 match
boxes containing these objects, based on their masses (which were, of
course, bound up with the box mass - the boxes could not be opened). With
about 20 of these, and drawing a histogram of the frequency of occurence
vs. the box mass (and it's left to the student to set an appropriate bin
width for the histogram), one can then begin to reason about potential
properties of the objects in the boxes. Of course, the more or less
uniform spacing of the peaks (with the odd "missing" peak - I spiked the
samples) suggests that the objects have the same mass and that the
observational differences had to do with differences of whole numbers of
masses, and probably the differences were due to differences of _one_
object (Occam's razor, etc.), and that it appeared that one could never
observe a single object (again, my spiking, just to provide a blind alley
or two in the analysis, and so that they couldn't figure out that there
was one object by shaking - even that was overcome another way - see
below). Anyway, the point of the exercise was to observe, indirectly
(i.e., via mass), the quantization effects in observations, and the
_detailed reasoning_ that would be required to sort it all out, including
comments on limitations (i.e. what were the assumptions being made in
analysis, insufficient data to say much about the probability distribution
- I had created something of an exponential decay in frequency with mass,
and so on), and also the determination of an object mass (averaging the
differences between the mean mass values of the peaks). It's a really
good exercise for scientific thinking and introducing the quantization
effect.
I was using machine nuts in match boxes. Interestingly, to keep the unit
mass far enough away from the box mass I used groupings of three nuts to
represent one object (which is a possible interpretation of the results
that is usually rejected based on assuming the simplest explanation is the
best).
Second, the students rarely got more than three or four charge values
during an experimental run. This was part of the point too! When they
first looked at Millikan's apparatus in a book it looks so simple, and
elegant (it is!) and easy to do (it isn't!). One function of the lab was
to show what a clever design (apparatus and procedure) the approach was.
Another function was to show just how hard it still was to get the
convincing results that Millikan did (in spite of a few 1/3 e and 2/3 e
values :-) ). There are several examples of wonderfully simply approaches
to experimental determinations of something or other, in which the
textbook can give students the idea that this removes any need for great
experimental care, and sometimes, a lot of grunt work. Cavendish's
experiment, the photoelectric effect, Franck-Hertz, and Rutherford
scattering are all such examples.
Finally, we pooled the class results and tried to make some sense of the
lot of them, now having better statistics. Interestingly, it was a great
discussion in which some groups of experimenters came under fire by others
for not being careful enough, and them defending their approach. It was
good science.