The biggest gaps are caused by the bumps near 50 and 63 Hz, and killing those would probably get you close to the 42 Mpc reported by Paolo (his calculation seems right). The real gain compared to the nominal curves, however, should come from lowering the noise floor between 50 and 200 Hz.
For the record, the peaks at ~60 Hz are calibration lines (see attached). Nominally, only the lines appear in DARM. However, there was an elevated bump during the timing issues (and when Bas made his plot); this bump is also elevated during our IB shaker tests. We have plans to investigate possible scattered light associated with these, but since nominally the bump is absent in Hrec, perhaps just the subtraction could be made better.
