A well-known cross-linguistic pattern in bound person paradigms is for zero marking to be restricted to third person forms, particularly third person singular subjects. This has been explained in terms of general language preferences for third person zeroes, possibly determined by principles of iconicity or economy. This paper discusses several diachronic phenomena that result in third person zeroes cross-linguistically, and argues that these phenomena pose two challenges for this type of explanations. First, several phenomena do not appear to be related to general language preferences for third person zeroes. Second, different phenomena are a result of different causal factors, meaning that third person zeroes are not amenable to a unified explanation. This calls for a source-oriented approach to third person zeroes and recurrent cross-linguistic patterns in general, one where these patterns are accounted for in terms of the properties of different diachronic phenomena that shape the pattern over time, rather than synchronic properties of the pattern in itself.