As we've got some large low frequency PCAL lines I had a look again at the how much of these lines appeared in OMC REFL and the DCPD to estimate of how much DARM signal is being rejected by the OMC, whether this is from alignment+mode-matching who knows. 

The ratio of transfer functions [REFL/PCAL] / [DCPD/PCAL] at 8.925 Hz and 11.575 Hz are both about 16-17%. The 8.925Hz line has a coherence of about 0.9 and 11.575 Hz 0.7 with H1:OMC-REFL_A_LF_OUT_DQ.

Late last year we also tried putting a 1Hz line in DARM and looking at the its ratio in AS_C and REFL (after rescaling REFL and AS_C to overlap the noise spectra). Back then we found REFL/AS_C ~ 24%. Quickly doing the same for the 8.925 Hz line we also see 15% in reflection.

These numbers are not so far away from what's needed to explain a good chunk of the mystery readout losses perhaps - 68912

The main uncertainty here I think is in what local oscillator field is the DARM signal beating with at OMC REFL, is it just beating with higher order mode junk light that amplifies how much we think is being reflected? If you believe the modulation depth tests (most recent one just before the ER) then 30% of the light at AS_C (calibrated as power going into HAM6) is carrier light... There is about 830mW in total going into HAM6 so 250mW of that is carrier. Only around 20mW of that is TEM00 light that transmits through the OMC on to the DCPDs, leaving 230mW of higher order junk at OMC REFL. Which is a lot of junk light to be suspicious of. It's unclear at the moment how much of this would overlap with the DARM signal rejected from the OMC but perhaps not unreasonable to assume there is some.