Although VLBI reveals structure in active galaxies on scales as small as a few parsecs, this is still likely to be several orders of magnitude larger than the scales on which jets are accelerated. We must therefore rely on theory, bolstered by indirect evidence, to deduce the behavior of plasma immediately surrounding the “central engine”. One hypothesis which has received much attention recently postulates that jets emerge from thick tori of gas which surround massive black holes. The low optical and X-ray luminosities of radio galaxies imply that tori in these objects would have to be supported by ion pressure, while tori in quasars and Seyferts may be supported by radiation pressure. Rees et al. have proposed that two parameters, α(the ratio of viscous stress to pressure) and (the ratio of the accretion rate to the value associated with the Eddington limit), are sufficient to determine the general characteristics of a torus, independent of the mass of the black hole. While existing observations of radio jets do not contradict the existence of such a homology, systematics involving the broad emission lines in quasars and Seyferts suggest that the homology may fail in these objects, at least insofar as conditions at ≲1 parsec reflect conditions in the torus. I report here on how the onset of nuclear reactions in a radiation-supported torus may break the homology, and regulate the properties of the broad line regions in quasars and Seyferts.