This is a bit of an aside, but as a methodological question, do you do your spectra as FFTs on weighted data? Or do you use multitaper methods? Although clearly trained in the former, with actual datasets, I tend to use the multitaper methods.