Transposons constitute much of the animal genome. While many transposons are ancient and inactivated, numerous others are intact and must be actively repressed. Uncontrolled transposons can cause genomic instability through DNA damage or mutations and must be carefully silenced in the germline or risk sterility or mutations that are passed on to offspring. In Drosophila melanogaster, 23–30 nt long piRNAs direct transposon silencing by serving as guides for Aubergine, Argonaute3, and Piwi, the three fly PIWI proteins. piRNAs derive from piRNA clusters—large heterochromatic DNA loci comprising transposons and transposon fragments. piRNAs are loaded into PIWI proteins via the ping-pong cycle which serves to amplify guide piRNAs. Loaded Piwi then enters the nucleus to transcriptionally repress transposons by establishing heterochromatin. Therefore, to silence transposons, transposon sequences must also be expressed. To bypass this paradox, the HP1 homolog Rhino (Rhi) allows non-canonical, promoter-independent, transcription of transposons embedded in heterochromatin. Transposon RNAs produced in this manner are “incoherent” and have little risk of being translated into transposon-encoded proteins required for transposition. This thesis focuses on understanding how piRNA clusters permit non-canonical transcription yet restrict canonical transcription. We found that although Rhi promotes non-canonical transcription in piRNA clusters, it also creates a transcriptionally permissive environment that is amenable to canonical transcription. In addition, we discovered that the conserved protein, Maelstrom, is required to repress promoter-driven transcription of individual, potentially active, transposons within piRNA clusters and allows Rhi to transcribe such transposon sequences into incoherent piRNA precursors.