Silver has unique optical properties for topical applications such as plasmonics. The two most widely used silver optical data sets are the Palik handbook compilation and that determined by Johnson and Christy. Unfortunately these are inconsistent making realistic modelling of the likely performance of silver in optical applications difficult, with modelling producing either highly optimistic or very pessimistic results, depending on application. By critical examination and duplication of the original experiments leading to the widely accepted literature values, we show that both data sets have drawbacks and conclude that there is a need for an improved data set for realistic simulation of experimentally obtainable properties.