To date, no historical study of signed languages has successfully applied the comparative method to identify the types of regular correspondences that conclusively establish cognate vocabulary. I explore theoretical and methodological reasons for the failure of the primary method in historical linguistics when applied to signed languages. I argue that this failure to demonstrate cognacy stems from three problems: 1) the prevalence of iconicity and indexicality in the lexicons of all known signed languages, 2) differences in the transmission of signed languages compared to spoken languages, and 3) the lack of cross-linguistic lexical databases using comparable representations.