Had a closer look at the data from the dither test in 6607, which apparantly showed much more astigmatism in the HWS magnification than is measured with every other technique.

It appears that there is a relatively larger cross coupling in the pitch/yaw signals seen in the HWS prism X/Y compared to the OPLEV in both the ITMX_cross_coupling and ITMY_cross_coupling dither tests. It is definitely worse for HWS-ITMX, but its hard to properly measure since there isn't quite enough samples in this dither test.

I went and reprocessed the HWS images over this time period, without removing tip/tilt in the images as is usually done in the HWS code. This way you can see the pitch & yaw dither in the wavefront images as a steep gradient. If you average the absolute OPD distortion over the duration of each dither test and overlay the images you see this for ITMX_avg_dither, ITMY_avg_dither. The lines of zero displacement (should?) define the horizontal/vertical axiis of the test mass suspension.

It appears the testmass pitch/yaw axiis are rotated in the hartmann images. This could explain the extra cross coupling seen, and maybe be related to the weird results for the astigmatism in 6607.

I have done some rough fitting and it looks like the axis in ITMX is rotated 6.74 deg CCW, and ITMY is rotated 2.01 deg. I get basically the same answer fitting for the rotation of the horizontal/vertical axiis.