It has been claimed that truly progressive harmony is unattested—all putative progressive harmony processes are reducible to (i) stem-control (e.g. Turkish), (ii) stress-control (e.g. Claro), or (iii) initial prominence-control (e.g. Tutrugbu). We show that sibilant harmony Santiago Tz’utujil (Mayan) disproves this empirical claim. STz’ exhibits progressive sibilant harmony that is divorceable from (i)-(iii). This has two analytical consequences. First, pure progressive spread must be representable in the phonology, regardless of one’s choice of phonological framework. Second, a particular diachronic pathway can be proposed that explains the relative rarity of the harmony process. In a nutshell, a perfect storm of grammar-particular and grammar-external facts conspired together to give rise to progressive sibilant harmony in STz’. Our proposal on the emergence of the process sheds light on possible sources of sound change and how learners phonologize a pattern in the input; i.e., how they may attribute a surface input pattern to the phonology of the input grammar, when the pattern was not in fact the output of a phonological process.