An Update on the Sproul Data

Back in January 2011 Eric Jensen of Swarthmore College, home of the Sproul Observatory, provided me with three stacks of papers from the late Dr. Peter Van de Kamp. These pages contained raw and processed astrometric data on epsilon Aur which. As you might recall, my initial post on this topic explained that there is a strong disagreement between the orbits implied by the Van de Kamp orbital solution and the more recent interferometric data from CHARA.

In the intervening months since my previous post Paul Hemenway and I went through the papers and figured out what they actually contained. Then I digitized a subset of the pages and checked the results for errors. In the last few weeks I've written some code to parse the data and start reducing it. Keeping in tradition with my other posts I'll break this lengthy process down into a few sub-posts over the next few days/weeks explaining what the data is, how it was obtained and measured, how the reduction process works, and what we can hope to expect.

In this post I'll focus on the first two items discussing the data and the measurement process. It's really quite interesting from both an instrumental and, even more so for me, a technical/historical perspective. I've been learning about long-focus astrometry as I've been doing this project so I'll end up providing some details. I'm hoping an astrometrist is reading this so a professional in this field can supply some comments and/or elaborate when they think it is necessary.

What is Astrometry

In short astrometry is the measurement of star positions and their motions (see Wikipedia's entry for more detail). There are many techniques to do this. You can use observations taken through telescopes with very accurate pointing, photographs of star fields taken over a long period of time, or even some interferometric techniques. Positions of stars are measured to incredible precision and their motions to the milliarcsecond level (i.e. 0.001 of an arcsecond, or 1/3,600,000 of a degree, or ~ 4.848 nano-radians) and getting this type of precision is very complicated (more on this during a later post).

What Are these Data

The data are a series of exposures of epsilon Aurigae and surrounding stars on photographic plates taken between 1938 and 1978 using the Sproul Observatory refractor. A more detailed account of the telescope can be found in Van de Kamp (1978 and references therein). My understanding of this process is that a plate of glass, coated with photographic emulsion, would be placed in a holder in the focal plane of the telescope and exposed to the starlight for a period of time. Photographic plates weren't cheap so it was common for there to be multiple exposures on a single photograph.

The surrounding stars are much dimmer than eps Aur so the observers used a cleaver trick. They used a rotating sector (think "fan blade") to periodically cover up eps Aur, thereby decreasing the number of photons reaching the photographic emulsion. The sector would have been off to the left in the above image.

Although there are 18 stars in the image, not all of them were used. eps Aur (labeled "π") in the image and four so-called "reference stars" (labeled 1, 2, 3, 5 in the image) were measured and their positions recorded.

How they were measured

Measuring the position of stars on plates seems fairly straightforward. Just get out a ruler and record the (x,y) positions, right? Almost so. The observers would have used a plate measuring machine. The best resource to describe these devices I've found is at the University of Virginia where they have a Virtual Museum of Measuring Engines. The data from Sproul were measured on a Grant measuring machine (similar to the one described in that article).

To measure a plate the operator would place the plate on the measuring machine and visit each image of each star in some particular order. Once the star was centered the operator (or a computer) would record the position of the star for use later.

My reductions of the Sproul data showed something weird. The plate scale I derived was 1 sproul unit = 9.34 arcseconds (asec), whereas Van de Kamp stated 1 mm = 18.87 asec. After reading up on the measuring machine, this difference was easily explained away in that the Grant machine has a resolution of 2000 steps / mm (rather than 1000 steps/mm) so they would measure the plates with twice the formal resolution of the image!

-------

In my next post I'll discuss how I am reducing these data.  Until then, clear skies!