🚀 go-pugleaf

When I was in college, the following experiment was among those 
required of us:  with the use of a watch, a microscope and two fiducial
lines on the lens, we had to time the descent rate of minute
particles of glass (microspheres).  The goal was to determine the
charge of a single electron from the speed measurements as the
spheres traversed a known distance in a known electric field.
In the absence of the electric field, the motion of the spheres
was somewhat brownian, with a slight bias for gravity.  But in
the presence of the electric field, the spheres reached a terminal
velocity that was a function of the number of charged particles
trapped on each sphere.  

For the sake of this discussion, the physics of the experiment
are irrelevant.  What is relevant is the technique by which
Milliken estimated the charge of a single electron.  The experiment
yields a wide distribution of velocities.  If all the spheres
had a single electron trapped upon then, then the distribution
would be centered about the velocity that corresponded to the
charge of one electron.  But there is no way of knowing how many
electrons are trapped on any sphere.  So, Milliken (and I ;-) 
collected many data points and tried to determine graphically,
by plotting a histogram, the charge of a single electron.
The lowest observed peak corresponded to the charge of a single
electron, and the other observed peaks corresponded to multiples
of the lowest, approximately.

Needless to say, Milliken was far more successful with this experiment
than was I.  But the question has come up again, in a slightly different
form.  Given a sequential listing, what algorithm can I use (program) to
determine the various means (or peaks), probability density functions,
and appropriate summary statistics?  I can spot the peaks manually,
of course, by simply plotting histograms and looking at them.  But I
would like to automate the process, since I might have to study
hundreds of listings.  References are most welcome, of course.
Thanks.

Tony Rizzo  (att.com!hogpb!rizzo)

(greetings Art.  How's the weather in CA? ;-)