Studying each component or equivalently the observations in individual spectral domains provides information on the physical processes at the origin of the components. It does not, however, provide a full picture in which it is possible to see how the gravitational energy released by accretion is distributed between the various cooling channels, nor does it allow us to describe the respective geometrical arrangements of the components or their physical relationships. Correlated studies across the complete electromagnetic spectrum are necessary for this research. 3C 273 is a prime source for these studies as it is bright in all bands and located close to the celestial equator. Both characteristics provide for ease of access with many instruments.
An early and surprising result was obtained by (Courvoisier et al. 1990) who showed that the UV light curve leads the radio emission by a few months. This result is confirmed by continued monitoring (Courvoisier 1997) and may be understood if the blue band flux is a signature of the accretion process (and hence of the energy release) and the radio (synchrotron) emission one of the cooling channels located at some distance from the central black hole. In this case and assuming that part of the accretion energy is carried along the VLBI jet (described in Sect. 6.2) and with its velocity from the central source to the location of the radio emission, the observed delay of approximately 0.4 year can be used to estimate the distance D at which the radio (22 GHz) flux is emitted:
where 
j
= 0.95 is the VLBI jet velocity divided by c and
cos 
j = 0.95 the
cosine of the jet angle with respect to the
line of sight
(Davis et al. 1991).
These data imply that the radio
emission is located some 4 light years from the central source along
the jet.
(Clements et al. 1995) have also performed a correlation analysis of the blue bump and radio light curves. They used photographic photometry from the Rosemarie Hill observatory taken from 1974 to 1992, radio data from the University of Michigan Radio Astronomy Observatory and data from the Algonquin Radio Observatory from 1966 to 1990. Cross correlating the optical and radio light curves does not provide a significant signal at any lag. This stems most probably from three factors. The blue bump light curve is undersampled and was obtained from B band observations rather than ultraviolet. Furthermore the radio light curve is at 10 GHz, less than that used in the preceding analysis. At lower radio frequencies, the light curves are smoother and the amplitude of the variations decreases (Teräsranta et al. 1992). This effect probably smoothes the variations in 3C 273 to the point at which the correlation with the optical variations is lost. It is interesting to note, however, that (Clements et al. 1995) and (Tornikoski et al. 1994) do find significant correlations between the radio and the optical light curves in several other objects. The optical light curves always lead the radio light curves. Typical lags are of the order of several months similar to those obtained in the case of 3C 273.
Using the radio and ultraviolet data, one may also wonder whether it is possible that 3C 273 is a mis-directed BL Lac object. Indeed 3C 273 has a strong synchrotron source emitting in the radio and millimetre domain and a superluminal jet. Both are characteristic of BL Lac objects. Were the blue bump and the emission lines overwhelmed by the synchrotron emission, one might well classify 3C 273 as a BL Lac type object. For this to be the case, the synchrotron component should, however, be boosted by a factor larger than 103 (Courvoisier 1988). The resulting radio flux would then be larger than 3 . 104 Jy, a highly improbable flux. It would thus appear that the presence of a strong blue bump and bright emission lines is an intrinsic difference between BL Lac objects and quasars like 3C 273 rather than due to orientation effects.
The relation of the UV with the X-ray emission is of considerable interest to test reprocessing models. One aspect of this correlation has already been discussed when the slope of the X-ray component was compared to the ratio of X-ray to UV photons. Cross correlating the IUE and BATSE light curves, one finds a very significant correlation peak at a lag indicating that the X-ray light curve follows the UV light curve by 1.75 years and no significant correlation close to zero lag (Paltani et al. 1998). Assuming that this result represents a physical reality rather than a chance occurrence of features in the light curves, (Paltani et al. 1998) conclude that the Comptonising X-ray emitting medium could be heated in a shocked region formed in a mildly relativistic wind at about 1 pc from the central source, in good agreement with the model proposed by (Courvoisier & Camenzind 1989) and described in Sect. 8. It is, however, also possible to apply models in which the Comptonising medium is located on the surface of the soft photon source (e.g. on the surface of an accretion disc). In this case, the flux correlation cannot have a physical meaning and, provided that the temperature of the plasma is known (e.g. from (Walter & Courvoisier 1992)), one deduces an X-ray spectral slope as a function of time assuming a variable optical depth in reasonable agreement with the existing data.
Looking at the correlations between the X-ray and the radio light curves, one finds no significant correlation at zero lag (Courvoisier et al. 1990), (Courvoisier 1998). This indicates that the X-rays cannot be due to a simple synchrotron self-Compton process in which the radio photons are scattered by the same electron population that produced them in the first place. The light curves now available indicate that the medium energy X-ray and radio components are correlated when the X-rays emission follows the radio by 2.2 years. This result is however, based on a dominant flare in the X-ray light curve, it remains then to be seen whether it proves solid with time.
Taken at face value the data presently available indicate that the UV flux leads all the other components. The typical delays are of the order of one or very few years. This result is in strong need of confirmation. It will, however, take many years of careful multi-wavelength observations to do so. This result would considerably strengthen the conjecture that the UV emission is a signature of the accretion and that the released energy is transported from the central regions of the gravitational potential well to the regions where it is radiated by relativistic or near relativistic flows.