A differential equation in which the time derivative of an unknown function also depends on the past information is mathematically formulated as a retarded functional differential equation (RFDE). The key to this formulation is the history function, which is a function that stores information about the history of the unknown function, and the (generally vector-valued) functional for the history function. It is usual to think of the history function as a continuous function on a closed and bounded interval $[-r, 0]$, however, there is a need to carefully consider the function space to which the history function belongs. In this talk, I will introduce a notion of mild solutions to linear RFDEs that allow for discontinuous initial history functions and explain how they can be used.