We examine the mathematical and physical significance of the spectral density sigma(omega) introduced by Ford [Phys. Rev. D 38, 528 (1988)], defining the contribution of each frequency to the renormalised energy density of a quantum field. Firstly, by considering a simple example, we argue that sigma(omega) is well defined, in the sense of being regulator independent, despite an apparently regulator dependent definition. We then suggest that sigma(omega) is a spectral distribution, rather than a function, which only produces physically meaningful results when integrated over a sufficiently large range of frequencies and with a high energy smooth enough regulator. Moreover, sigma(omega) is seen to be simply the difference between the bare spectral density and the spectral density of the reference background. This interpretation yields a simple "rule of thumb" to writing down a (formal) expression for sigma(omega) as shown in an explicit example. Finally, by considering an example in which the sign of the Casimir force varies, we show that the spectrum carries no manifest information about this sign; it can only be inferred by integrating sigma(omega). (C) 2011 American Institute of Physics. [doi:10.1063/1.3614003