Recreating the Millikan Experiment: Analysis and Error Considerations

[randybryan]

We've had to recreate the famous Millikan oil drop experiment in our labs and taken values of the velocities of oil drops under the influence of gravity and charged capacitor plates and the balancing voltage required to stop the droplet from moving up or down.

We took 120 results and they look promising. On a histogram of charge vs frequency there appear two peaks that the data cluster around. The difference between these peaks should give an indication of the unit of charge that we're trying to determine. Similarly on a graph displaying the calcculated radius of the droplet vs charge, there are distinct clusters centrered around a common charge value (yet to be determined).

I'm now wondering how to go about looking into error. There are so many ways to approach the analysis and I was wondering if anyone who may have performed the experiment themselves could comment on what they think is the best method. Should I use a Gaussian approximation to the histogram?

Would be very grateful for advice and can elaborate on any details you need

 

[Pythagorean]

with 120 points you could take the mean and deviation of each cluster.

 

[Bob S]

Don't you have more than one variable: voltage, drop size, and charge? Then you derive a probability plot of drop charge, from which you extract the peak separation? What is frequency; do you mean probability or Hz? FYI: When I did the experiment, I got a value of 0.8 x 10^-19 Coulombs.

Bob S