Nuclear run-on is a technique used to estimate the relative rate of the transcription of a given gene, as opposed to steady-state level of the mRNA transcript (which is influenced not just by the transcription rates, but by the stability of the RNA). This method is based on the assumption that the highly-transcribed gene should have more molecules of RNA polymerase bound to it than will the same gene in the less-active state. If it is properly prepared, isolated nuclei will continue to transcribe genes and incorporate 32P into RNA, but only in those transcripts which were in progress at the time the nuclei were isolated. Once the polymerase molecules complete transcript they have in progress, they should not be able to re-initiate the transcription. If that is true, then amount of the radiolabel incorporated into a specific type of mRNA is theoretically proportional to the number of RNA polymerase complexes present on that gene at the time of isolation. A very difficult method, rarely applied appropriately from what I understand