Abstract
Traditional philosophy of science believes that scientists can achieve agreement on every experimental result provided it can be replicated in an appropriate way, that is, reproducible with the same experimental arrangement and procedure. By analyzing the role of skills in experiment appraisal, I explain why in fact scientists do not always have consensus on experimental results despite their replication attempts. Based on a detailed analysis of a historical case, I argue that experiment replications inevitably involve a processor skill transference, which is frequently not articulated in linguistic discourses. Hence, it is very difficult to make identical replications if experimental reports are the only resources. Furthermore, I argue that, because transferred skills have to be integrated with scientists' prior experience, skill-transference is sensitive to contextual factors, which can prevent scientists from reaching consensus on experimental results by influencing the effectiveness of communication in experiment appraisal.