To test how previous findings from industrial and academic contexts apply, and to grant the replicability of our experiment, we focused on OSS systems forour investigation. We examined over 30 mature OSS systems claiming to do mandatory, continuous code review as possible study candidates. Unexpectedly, only two projects adhered to their claim. For these two systems, we manually classified more than 1,400 review changes into a defect categorisation, which we validated in an interrater-reliability study including two developers of those systems. We classified each change manually with the help of an Eclipse plug-in that we developed. Using the same setup,in a second step,we classified the trigger for the change.