Eran Tal (Bielefeld) gives a talk at the MCMP Colloquium (12 June, 2013) titled "A Model-Based Epistemology of Measurement". Abstract: The epistemology of measurement is an interdisciplinary area of research concerned with the conditions under which measurement and standardization methods produce knowledge, the nature, scope, and limits of this knowledge and the sources of its reliability. A primary goal of such studies is to better understand the ways in which theoretical and statistical assumptions about a measurement process influence the content and quality of its outcomes. Such assumptions often involve idealizations, that is, intentional distortions of aspects of the measuring instrument, measured object and environment that appear to threat the accuracy and objectivity of measurement. Here I argue that the opposite is the case: idealization is a necessary precondition for obtaining accurate and objective measurement outcomes. A measurement outcome, I submit, is a value range assigned to a parameter in a model in a way that allows the model to coherently predict the final states (‘indications’) of a process. Idealizations are necessary for identifying the measured parameter with a particular object, for distinguishing genuine effects from errors and for comparing measurement outcomes to each other. These claims are exemplified with a study of the contemporary evaluation and comparison of atomic clocks across national metrological laboratories. Building on these insights, I conclude by highlighting the promise held by model-based approaches for further research in the epistemology of measurement.