Due to an error in typesetting, the reference numbers in the text of Matthews et al. (2004) do not match those in the list of references. The correct numbers are given in the Erratum, which we repeat here for convenience: References 1 to 26 should be, respectively: 1, 10-16, 2, 3, 17-26 and 4-9.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

We follow the usual definition that the lifetime of a damped oscillator is the time for the amplitude to decay by a factor of e.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...Gilliland et al. (1993)

Note that all four power density spectra in Fig. 2 were scaled in the same way, such that a sinusoid with an amplitude of 1 ppm observed for 10⁶ s gives a peak in power density of height 1 ppm²/ $\mu$ Hz. This explains the factor of two between the VIRGO power density in the figure and that plotted by Pallé et al. (1999), who used a different scaling convention.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.