Visual modeling notations are ubiquitous in software engineering. However, the visual (aka concrete) syntax of such notations is currently designed in a largely ad hoc and unscientific manner, without reference to theory or empirical evidence. This is problematic since human understanding of models relies as much on syntax as semantics. The aim of this tutorial is to establish a scientific basis for visual notation design: to help it progress from a "craft" (as it currently exists) into a design discipline. It presents a set of principles defined by Daniel Moody and called The "Physics" of Notations. These are evidence-based principles for designing cognitively effective visual notations: notations that are optimised for human communication and problem solving. Each principle will be introduced and illustrated with examples and hands-on exercises. The students will be involved in analysing and re-desiging existing notations such as UML, BPMN and i*.