Lessons in Applied Data Archaeology

I'm teaching our upper-level lab course this term, where I do a two-part experiment on laser spectroscopy. The first part is to calibrate the free spectral range of a homemade Fabry-Perot interferometer, and the second part is to use that Fabry-Perot as a frequency marker to calibrate a diode laser scan across the rubidium hyperfine spectrum, allowing a measurement of the Rb ground-state hyperfine splitting.

That's a bunch of jargon, the details of which don't really matter. What matters is that this is a lab that involves scanning the frequency of a particular laser through some range of frequencies, and on Sunday, I blew up that laser. I forgot that the current supply had been used for something else since last spring when I did the lab, and thus the current limit was raised above the maximum safe level for that diode. When I started turning the current up, I expected the limit circuit to stop me from blowing up the diode, and it didn't. When that happens, *poof*, a laser becomes an LED.

Which leaves me in a tough spot, because that laser was the only one that hit the right wavelength for this lab. And I have so much other stuff going on that I don't have time to track down another working diode.

Which means that the students in this year's lab will be learning another useful experimental skill: data archaeology.

We use the same Fabry-Perots every year, so the FSR should be about the same, and we use the same computer for data analysis purposes, so there are several past classes' worth of data files on the machine. Accordingly, I just copied an old data set into the folders for this year's class, and they'll use that to do the analysis. I'll still have them come down and see the apparatus, and talk about how they would've acquired the data, and then they'll be turned loose on the old dataset to do the usual analysis.

Of course, the fact that this is old data from another class means that it's incredibly poorly documented. Any notes about the settings used for the data files were presumably kept in the notebooks of the lab students, and taken with them after graduation. Which means that the students get to figure out what the files are by opening them up and looking at the data directly.

This is a surprisingly common occurrence, even for professional scientists. I don't think I've ever written a paper without needing to do a little data archaeology-- there are always a couple of files that didn't get written down, or some parameters that need to be reconstructed from sketchy notes in the lab notebook, or some exceptions that need to be figured out from odd little shorthand in the margins. Real experimental data is taken over a period of weeks or months, and writing it up often involves some reconstruction of the experimental situation from incomplete information.

So, while this is sub-optimal in lots of ways, there's a sense in which it's providing a useful lesson. At least, that's what I'm telling myself between self-administered boots to the head for frying the stupid diode.