This chapter presents the flaws in the Council of Science Editors (CSE) guidelines, through presenting basic ways of classifying scientific images that are implicit in the academic literature. It aims to provide a robust philosophical analysis of the production and content of scientific images and foster an appreciation for their boundary object role, upon which better norms around their publication might be devised. The chapter introduces the concepts 'trading zones' and 'boundary objects'. It discusses the three distinctions for analysing scientific images. Three main distinctions: the distinction between visual languages and pictures, the distinction between machine-produced and human-made images and the distinction between the epistemic role of generic exemplars and the epistemic role of data. These distinctions indicate the variety of ways in which scientific images display their content and the character of their epistemic power and reveal the limitations of the CSE guidelines.