The secondary target for the 15th running of the Whole Earth Telescope, DQ Herculis, is an eclipsing binary star that exhibits "flickering" in its light curve, as well as extremely coherent oscillations with a period close to 71 seconds. Our object is to study the character of these rapid oscillations, but the eclipse, which occurs every 4h 39m, and the "flickering" --- up and down excursions in brightness with time scales of 5 to 30 minutes --- dominate the light curve; the rapid oscillations are superimposed on them, and are therefore very hard to study in vivo.

Darragh O'Donoghue has come up with a simple procedure to extract them, and the accompanying diagrams show the result for just one of the many runs already in hand, to which this magical procedure will be applied. Figure 1 (below) shows the extracted oscillations in the upper panel and the (smoothed) light curve in the lower. Notice that the size of the oscillations is fairly constant around the orbit, but falls to zero during the eclipse (two of which are shown).

The procedure involves two basic steps, as follows:

1. After normal reduction of the light curve to identify sky readings and to remove bad points from flashlights, bugs crawling over the photometer apertures, etc., an exact copy of the curve is made, and a long boxcar smoothing process is applied to it --- in this case 142 data points, or 284 seconds. The resulting smoothed light curve (bottom panel in Figure 1) is then subtracted from the original, flattening it but leaving almost unchanged the more rapid processes: the oscillations, and the "flickering".
   
   
2. An exact copy of this "flattened" curve is made, and smoothed a second time with a boxcar covering 36 points (72 seconds). It really should be 71 sec, but this is the best we could do on short notice, and it is probably Close Enough for Government Work. This removes the oscillations, leaving behind the (smoothed) variations from the flickering process, which can then be subtracted from the first flattened curve --- removing the unwanted variations and leaving behind the rapid oscillations, ripe for further study.

Figure 2 shows the same oscillations as Figure 1, only stretched out over more real estate and supplied with an expanded Y axis so they can be seen better. Figure 3 shows another example of extracted oscillations, from a beautiful set of data taken at Pic du Midi.

We must be careful to calculate just how this double filtering process will affect them before we apply it to all the data in hand, but once all the runs have been joined together we should have a data set unprecedented in its potential. We can expect surprises.

FIGURE 1

(click on the image to see a larger version)

---

FIGURE 2

(click on the image to see a larger version)

FIGURE 3

(click on the image to see a larger version)