The article presents the findings of a survey of the imprecation phrase ma yi fei shi 麻夷非是 and its variations, as written (using brush and ink) on several thousand excavated covenant texts (mengshu 盟書) from Houma 侯馬 and Wenxian 溫縣. I argue that the findings support Zhu Dexi 朱德熙 and Qiu Xigui 裘錫圭's analysis of the phrase as mi yi bi shi 靡夷彼氏 “Wipe out that shi” (shi, I suggest, referring to the covenantor and his direct male descendants). Through comparison of scribal hands, I demonstrate that those variations which do not fit this analysis were produced by a small number of scribes and, in almost all cases, can be shown to be errors. I conclude that such variations are generally unreliable and do not require us to reject Zhu and Qiu's analysis. These examples suggest that formulaic, possibly archaic, stock phrases, such as this imprecation, were liable to be misinterpreted, even during the period in which they were in use. Identification of scribal hands and scribal errors was essential to this analysis, demonstrating not only the importance of this methodology in such research, but also the potential value of these particular materials for furthering our understanding of scribal habits and text reproduction in early China.