Blue, Green, and Red Bumps in AGN

I show that the summed spectral energy distribution (SED) $L(\nu)$ of any extended blackbody radiator will scale in a predictable way if all parts of the body change in temperature by the same factor $X$, such that $L^\prime(\nu) = X^3L(\nu/X)$. This should for example apply to accretion disks around black holes, where $X$ is relative accretion rate, or external heating rate, but will not apply to changes in black hole mass. I summarise evidence that AGN optical-UV SEDs become progressively redder with decreasing luminosity, and show that the trend in colour versus luminosity shown by Mushotzky and Wandel (1989) is matched extremely well by taking a template high-luminosity SED and scaling it in the manner described above. This agreement is striking because it involves {\em no adjustable parameters}. The agreement breaks down at low luminosities because of stellar contamination and reddening. I then consider the colour changes of an individual AGN (NGC 5548) during luminosity changes, which according to the popular X-ray reprocessing model, should follow the scaling law well. However the observed changes are clearly {\em not} consistent with the simple scaling prediction. Instead, these colour changes are quite well explained by the mixing of a constant red component and variable blue component. Overall, there is then strong support for the ideas (i) that AGN optical-UV SEDs arise from accretion discs, (ii) that accretion rate plays a significant role in the very large range of luminosity seen in AGN, and (ii) that the inner regions of AGN vary independently of the outer accretion disc.