The requirement for highly accurate timekeeping is increasing as technology develops in the spheres of navigation, geolocation, internet communications and financial transactions. Currently, these activities rely on a globally synchronised coordinated time which is kept precise by signals from large, laboratory based atomic clocks. However, these signals can be corrupted or jammed, and in a geopolitically uncertain environment, a reliable local alternative is required. Chip scale atomic clocks provide a portable solution, but currently available devices rely on complex vapour and laser-based technologies which are expensive, consume relatively large amounts of power and have reliability issues in microfabricated form. A chip scale device based on radio frequency EPR with endohedral fullerenes could solve complexity, cost and reliability issues provided it did not sacrifice accuracy. The work in this thesis builds upon previous experimental results, demonstrating some of the improvements required to establish the technological basis for such a clock. An improvement in resonator quality factor is shown to yield a five-fold improvement in signal to noise ratio, with which the clock transition signal of 15N@C60 is acquired more clearly than was previously possible. The low and zero field transitions of 15N@C60 are successfully mapped, yielding a precise value for the hyperfine constant and demonstrating the potential of a single resonator to respond to multiple frequencies. This offers reduced complexity for future microfabrication and provides a proof of concept for a method of field stabilisation. These results are discussed in the context of a functional clock and suggestions made for future work toward its realisation.