4 Optical observations and data reduction

Ground-based optical observations were made in support of the ISO observations. Spectroscopic monitoring was carried out with the 1.8-m Perkins Telescope of the Ohio State and Ohio Wesleyan University at the Lowell Observatory, the 1.0-m telescope of the Wise Observatory, the 2.6-m Shajn Telescope of the Crimean Observatory, and the 1.5-m Mt. Hopkins Telescope of the Harvard-Smithsonian Center for Astrophysics (CfA). A detailed log of the spectroscopic observations can be found at the International AGN Watch website.

In addition, CCD photometry was made on the 1.0-m telescope of the Wise Observatory. The flux was measured using a fixed aperture of radius 7'', under seeing conditions of 2-3''. The B, V, R, and I instrumental magnitudes are listed in Cols. (3)-(6) of Table 4. They have not been scaled to any standard system.

The spectroscopic data were processed by the individual observers in standard fashion for CCD frames. However, the standard astronomical flux calibration techniques based on determining the instrument response function from observations of standard stars are only accurate for AGN spectrophotometry at about the 10% level even under ideal observing conditions. We used the standard stars only for relative calibration, and employed different calibration techniques for absolute calibration: the data from Wise Observatory were calibrated following the method described by Maoz et al. (1994). The data from Ohio, Crimea, and CFA were calibrated through scaling through [O III] $\lambda5007$ flux that was measured on five photometric nights, at F([O III] $\lambda5007$) = $(1.512\pm0.096) \times$ 10$^{-13}\,$ergs s^-1cm^-2.

On these calibrated spectra we measured the continuum flux by averaging the flux in a 10Å wide band centered at 5105Å in the rest frame of Mrk 279 ( $F_{\lambda}(5100\,\mbox{\rm\AA})$). The H$\beta$ line flux has been measured by linear interpolation between rest-frame wavelengths $\sim$4765Å and $\sim$5105Å, and line integration between 4770Å and 4935Å. The long-wavelength cutoff of this integration band misses some of the H$\beta$ flux underneath [O III] $\lambda4959$, but avoids the need to estimate the Fe II contribution to this feature and still gives a good representation of the H$\beta$ variability. We did not correct for the narrow-line, which is expected to be constant.

As the measurements made from the spectra are subject to systematic differences between the four data sets used, we applied the prescriptions of Peterson et al. (1999) to intercalibrate the data sets, and correct for aperture effects. The final continuum $F_{\lambda}$(5100Å) and H$\beta$ emission-line fluxes are given in Table 5. The spectrophotometric and photometric light curves are shown in Fig. 2. Using the results of Granato et al. (1993), we estimate that stars contribute for $1.2 \pm 0.6$ mJy or $20\pm 10$% to the mean reddening corrected 5100Å flux.