This paper provides definitions for some desirable properties of voting systems, including auditability, ballot secrecy, incoercibility, usability and accessibility. In the context of these desirable properties, it defines the class of end-to-end independently verifiable (E2E) voting systems that provide high levels of auditability, without requiring voters or election observers to trust polling machines or election officials. It provides examples of E2E systems, and compares their auditability properties to those of other voting system classes. Finally, it presents areas for further research in auditable voting systems.