I analyzed the data measured by Takahashi-san using additional 300-kg masses.
The gain and coherence are shown in the first and second figures, respectively.
The green and blue dots represent the measured vertical and horizontal data.
The green line indicates the modeled vertical gain, assuming the same compensation factor as in klog:31104.

Both transfer functions maintain the isolation ratio below 0 dB between 60-100 Hz, without any significant peaks.
However, we observe two peaks in the vertical data below 50 Hz, where only one peak should be present.

Additionally, the coherence was unexpectedly low, even around a few tens of Hz in this measurement.
It is possible that some factors contributing to this low coherence may have resulted in the additional resonant structures.