The synthetic response to a gravitational wave source in position 43 of Figure is shown in Figure . This figure shows a set of spectra weighted to the directions shown in Figure by match filtering against the predicted gravitational wave response. Each trace number corresponds to a direction shown on the grid in Figure . The spectra peak is at 0.25 milli-Hz, and the maximum amplitude is at trace 43 as expected, but the definition of the direction is poor.
The directionally weighted spectra for the raw data without gaining, earthquake removal or tapering is shown in Figure . The spectra are generally noisy, and no strong response similar to that in Figure is seen, but there is a promising weak event near 0.244 milli-Hz.
The original data then had a taper applied to the nonzeroed portions to remove the effects of the sharp truncations. This result is shown in Figure . Notice that the spectra are still noisy, indicating that the earthquake and other noise are still dominating. The event at 0.244 milli-Hz is clearer, but still not convincing. When the 1986 spectra in Figure is compared to the 1985 spectra in Figure , there is no similarity in the spectra at 0.244 milli-Hz. The event is then rejected as a possible gravitational wave signal. Similar events were found in other years and at other frequency ranges.
To reduce the effect of the noise caused by the high amplitude earthquakes and ocean noise, a gain was applied to the original data. This equalized each station's amplitude, reduce the ocean noise, and attenuated the earthquake events. The directional spectra produced from these data are shown in Figure . Notice that the event at 0.244 milli-Hz is now much weaker. The amplitude differences between the rings of Figure are much less pronounced.
Next, the earthquake noise was removed by the process described previously, followed by the gain and taper processes. The spectra are now flatter and the event at 0.244 milli-Hz is even weaker. For comparison, the same frequency range is displayed in Figure for 1985.
Finally, an interesting effect is seen in the low frequency part of the directionally-weighted spectra. In Figure a series of events are seen spaced at 0.012 milli-Hz spacings. These correspond to even divisors of a 24 hour period.