3.3 Luminosities
The issue of whether QSO redshifts are of cosmological origin was unambiguously settled by the work illustrated in Figure 5 (particularly the two leftmost panels). When sensitive enough observations are made, the galaxies in which the quasars reside can be detected. The relatively faint emission from the surrounding stars, gas and dust is called quasar fuzz. Since quasar fuzz is clearly emission from distant galaxies, each with the same redshift as the quasar they contain, the redshifts of quasars must be cosmological.

In each case in Figure 5 the quasar itself was overexposed, so that very faint emission surrounding the central point source could be detected. In each case shown in Figure 5 faint nearby emission was discovered from stars in a galaxy. (Top left) PG 0052+251 lies at the centre of a normal spiral galaxy. (Top centre) IRAS 04505–2958 (the central source) has apparently recently collided with the spiral galaxy whose remains lie towards the bottom of the image. The distance between the quasar and the galaxy is one-seventh the diameter of our Milky Way. (Top right) The plumes of emission are from dust and gas which PG 1012+008 has apparently captured in a collision with a nearby galaxy. (Bottom left) PHL 909 is at the centre of a normal elliptical galaxy. (Bottom centre) The quasar PG 1012+008, at the centre of the image, is merging with the galaxy whose core is the bright object just below it. Wisps of dust and gas show material which is being pulled away from the galaxy by the quasar's gravity. (Bottom right) IRAS 13218+0552 appears to be at the centre of two galaxies which have merged.
Once we accept that the redshifts of quasars are cosmological, there is no way to avoid the conclusion that they have huge luminosities. The emission from the quasar PG 0052+251, shown in the top-left panel of Figure 5, is clearly brighter than that of the entire surrounding host galaxy, even though in this image the quasar itself is overexposed.
The observation that the active nuclei of even the closest Seyfert galaxies appear as unresolved point sources of the light immediately suggests that the luminosity of an AGN is generated within a volume which is small compared to the size of a galaxy. More stringent limits on the size of the emitting region in AGN arise from considering their variability (i.e. how the luminosity changes with time). In some sources the luminosity changes significantly over a few days. This means that the time, ∆t, for light to travel across the entire source must be only a few days, because otherwise the changes in luminosity would be smoothed out by the delayed arrival times of the photons from the more distant regions of the source. This can be expressed mathematically by the general requirement that

where l is the size of the emitting source and ∆t is the timescale for observed variability. Using this to work out the size limit corresponding to a light travel time of a few days we have:where we have adopted ∆t = 10 days, converted this into seconds, and used an approximate value for the speed of light: c≈3 × 1010 cm s−1. Evaluating, and retaining only 1 significant figure, we have
which can be converted into length units more convenient for astronomical objects: