The Birmingham Solar-Oscillations Network (BiSON) consists of 6 semi- and fully-automatic stations dedicated to the collection of full-disc helioseismological data (Chaplin et al. 1996a). Full details of BiSON are given in Table 1 (click here). There are two stations in each 120-degree longitude band. Moving successively east in longitude: Sutherland and Carnarvon are in band 1; Narrabri and Mount Wilson in band 2 (Mount Wilson sits on the band 2-band 3 boundary); and Las Campanas and Izańa are in band 3. All the sites lie at moderate latitudes.
| Site | Latitude | Longitude | Altitude | Commissioned | Station |
| name | (deg N) | (deg E) | (m) | (year) | type |
| Tenerife | +28.30 | -16.50 | 2368 | 1975 | A |
| Carnarvon | -24.85 | +113.75 | 10 | 1985 | B |
Mt. Wilson | +34.13 | -118.07 | 1742 | 1992 | A |
| Sutherland | -32.38 | +20.82 | 1771 | 1990 | B |
| Las Campanas | -29.02 | -70.70 | 2282 | 1991 | B |
| Narrabri | -30.32 | +149.57 | 217 | 1992 | B |
A: mirror-fed, semi-automatic system; B:
equatorially-mounted, dome-based, fully-automatic system.
System moved from Haleakala, Hawaiian Islands, 1991.
The network spectrometers rely on the technique of resonance
scattering to isolate two narrow passbands in the blue and red wings
of a solar photospheric absorption line, formed by potassium atoms in
the near infrared (
). The normalized difference between
the measured blue and red intensities serves as a sensitive,
near-linear measure of the line-of-sight velocity of the source. The
BiSON instruments do not image the solar disc - the data collected
are therefore sensitive to those solar modes possessing the largest
spatial scales, i.e. for 
. The radial wave
functions of some of these modes penetrate to the core of the Sun.
The preparation of calibrated velocity residuals from Doppler velocity data collected by each of the network spectrometers, as described in Paper I of this series, is only the first step in the analysis of solar-oscillation data. In order to extract scientific results from the data, we need to combine the residuals into an extended time series, transform the time series into the frequency domain, and measure the mode characteristics in the spectra so obtained. This paper describes our approach to these tasks. In addition, we address the treatment of multi-station data overlaps between stations. Our discussion here concentrates on the signal-to-noise over the 5-minute region. Fossat (1992) considered a deconvolution treatment aimed at optimizing the signal-to-noise characteristics at low frequencies.