8 Description of the database  

The overall database is illustrated in Fig. 1 by a time versus frequency representation of all observations currently in the database and in Fig. 6 by the average spectrum of 3C 273. The observations presented here are publicly available on the WWW at the address: http://obswww.unige.ch/3c273/. Special care was taken to have a homogeneous and clear database, which can be easily understood and used. The user should however be aware that the presented light curves often contain observations at different (but close) wavelengths. To facilitate the selection of observations within the light curves, each individual measurement is always characterized by the epoch of observation, the frequency, the wavelength, the flux, the flux uncertainty, the observatory that made the observation and the literature reference of previously published observations.

The epoch of observation is given with an accuracy of up to one hour in universal time (UT). It is both expressed by the year (yyyy), the month (mm), the day (dd) and the hour (hh) coded in the number "yyyymmdd.hh'' and in fractional years of same duration (365.25days). The fractional year "Date'' is given with an accuracy of 0.0001year, which is about one hour (1day $\simeq$ 0.0027year). It was always derived from the Julian date (JD) by the following relation:

$$\mbox{Date}=2000.0+(\mbox{JD}-2\,451\,544.5)/365.25.$$ (3)

If an observation lasted more than one day, the epoch is set to the middle of the observation period. If "yyyymmdd'' finishes by two or four zeros, it means respectively that the day or the month of the observation was not known to us. In this case, the value of "Date'' was calculated for the middle of the month or of the year respectively. For a few radio catalogue data where we could not find any indication about the date of observation, we fixed the epoch to the year of publication, since it is a way to identify them.

We chose to give both the observed frequency and the observed wavelength at which the flux measurement was made. The frequency "Nu'' ($\nu$) is always expressed in Hz and the wavelength "Lambda'' ($\lambda$) is always expressed in Å. The flux density "Fnu'' ($F_{\nu}$) and its uncertainty "Delfnu'' ($\Delta\,F_{\nu}$) are always expressed in Jansky (1Jy=10^-23ergcm^-2s^-1Hz^-1=10^-26Wm^-2Hz^-1). Upper limits are not considered in this database. A zero value for "Delfnu'' means that the relevant uncertainty is not known to us. The observatory, the telescope site, or the satellite that made the observation is indicated by a name or an abbreviation in the column labeled "Obs''. For published data, the reference "Ref'' is given in an abbreviated form (initial letters of the three first authors' surnames and the year (yy) of publication). Instead of the "Ref'' column, the UV light curve files contain two other columns "Spectrum'' and "Flag'' containing respectively the INES spectrum identification and our general quality flag defined in Sect. 6.

The average spectrum shown in Fig. 5 and all individual IUE spectra are given in one file per observation in ASCII format. The files are simply the ASCII equivalent of the FITS files from the INES catalogue and the original headers are also provided. The date of observation and our general quality flag can be found in the UV light curves.

The original X-ray spectral fit parameters, which we used for constructing the light curves are also provided in a file. This file contains for each observation the epoch of its start and sometimes of its end, the energy range E₁-E₂ of the fit, the photon spectral index $\Gamma$=$\alpha$+1, the flux normalization at 1keV, $F_{\mathrm{\,1keV}}$, in units of 10$^{-2}\,\mbox{photons\,cm$^{-2}$\,s$^{-1}$\,keV$^{-1}$}$ (which corresponds to 6.6261 $\mu$Jy) and the associated uncertainties $\Delta\,\Gamma$ and $\Delta\,F_{\mathrm{\,1keV}}$.We also indicate the confidence level of the quoted uncertainties in $\sigma$ (1$\sigma$=68%, 1.6$\sigma$=90%, 2$\sigma$=95%), whether the fit was performed with a free value of the hydrogen column density $N_{\mathrm{\,H}}$ or not, the instrument that made the observation and the reference.