Presentation #236.05 in the session HAD II - Centennial of an Eclipse: The 1922 Expedition that Clinched the Case for General Relativity.
Although careful preparation is desirable, it is not always possible. My personal observatory, Alpaca Meadows Observatory (44.7909, -122.6096) located in Lyons, Oregon, was less than a kilometer from the center line. If offered electric power and Internet access. When Toby Dittrich proposed performing the Eddington Experiment and then received funding, we raced to prepare for the event using equipment already on hand: a TeleVue Genesis 100 mm f/5 refractor, an SBIG STT-8300 chosen for its rapid image readout, and a Celestron AVX equatorial mount. By June the equipment was set up in a compact roll-off shelter. Testing revealed that the Genesis needed to be stopped down to 80 mm, and its wobbly focuser had to be preloaded with springs. We determined the exposure time for the eclipse by imaging earthshine on the crescent Moon. To provide an astrometric reference field, we planned to slew the telescope west after mid-eclipse. Because there was no time to automate the imaging sequence, the students and I worked out a system where, working in alternation, they performed the necessary steps. Eclipse day was clear and the sky was smoke free. The students performed their roles perfectly. We obtained 23 images of the eclipsed Sun — rotating through 0.6, 1.0, and 1.6 second exposures — and 10 reference images. The telescope was immediately capped, the shelter rolled on, and we collected 500 dark frames. We made flat-field frames that evening. Six months after the eclipse, I shot reference images of the eclipse field at the same hour angle as on eclipse day. After calibrating the eclipse and reference images, the corona gradient was removed using wavelet filtering. The eclipse images contain 60 stars, of which 20 gave reliable centroids. Unfortunately, the plate scales of the eclipse frames, eclipse reference frames, and six-month eclipse field frames did not exactly match. As had been the case for Dyson a century earlier, which data to trust was a judgment call. We characterize the result as 1.7 ± 0.8 arcseconds deflection at the solar limb. Later analysis showed star positions were affected by severe turbulence on eclipse day. Thanks to the two students, Abraham Salazar and Jacob Sharkansky, Toby Dittrich, and especially Donald Bruns for his assistance in the reduction of the data.