Using their power from that catalog, their equation constant "C" derived from their SOX-E ballast specs (not gonna be the same for SOX ballast but hopefully close enough), their total power factor (also taken from SOX-E specs, again not gonna be the same but hopefully close enough) and the rest of their specifications put into a system of equations following this structure:
ZBal = (C * √(OCV^2 - VLamp^2)) / ILamp,
(VLamp)(ILamp)(PF) = PLamp
where VLamp is x and ILamp is y, I arrived at the following approximations:
26W SOX-E on 480VOC 35W SOX ballast:
60V, 0.61A
36W SOX-E on 480VOC 55W SOX ballast:
79V, 0.6A
66W SOX-E on 480VOC 90W SOX ballast:
86V, 0.94A
91W SOX-E on 650VOC 135W SOX ballast:
101V, 1.13A
131W SOX-E on 650VOC 180W SOX ballast:
188V, 0.92A
These are values calculated for the reference ballasts, which are linear impedances at 480 and 650VOC respectively. As is the case for all lamps but especially for LPS, these reference ballasts are usually never what is actually produced. These values I have calculated are wrong until proven correct. Again if anyone has the voltages or currents for these lamps on SOX ballasts please do let me know.
similar data to the philips catalog linked above but this time with actual current and voltage values can be gleaned on page 40
https://www.lamptech.co.uk/Documents/Brochures/Thorn%20-%20Brochure%20-%20D%20-%20Discharge%20Lamps%20-%201989%20UK.pdf(also keep in mind ignitor/CWA gear for SOX lamps is only a relatively recent thing, and only really applicable to the UK, in older times and in pretty much every other domain, its always been a leakage reactance transformer of the same OCV and general impedance as that of the reference ballast setups, so the 480V and 650V OCV values are still very applicable I would not be so quick to disregard them)