We installed a trillium compact with a reftek 130a at DSNC to test polarity. I got most recent dataless for DSNC from SIS. I modified dataless from an Arkansas temp station that used a trillium compact and reftek for station CATNT, the temporary at DSNC. I used sac to transfer from CATNT to the DSNC reponse using this command: transfer from polezero subtype CATNT.HHZ.ET.00.pz to polezero subtype DSNC.EHZ.ET.00.pz freqlimits 0.01 0.02 40 50.

I compared data for both a local event and a teleseism.

Plots follow for the local event, a magnitude 2.4 on August 12 near Cedar Rock, NC. Event page

Vertical Channels look reversed

North Channels look correct

East Channels look correct

Plots follow for the teleseism, a 6.4 on August 12 on the north shore Alaska. Event Page

Vertical Channels also look reversed

North Channels look correct

East Channels look correct

The GFM channels are less certain.

John's email from August 14, 2018: "As I understand it the problem with the ISIS4 canisters was that with S-13's the output is reverse of other seismometers and Greg and his partner forgot this when they were assembling them - except for the last one when Greg caught the problem S/N 14. We reversed the input wires in all of the canisters not remembering that GFM still had an L4-3D installed. Chris and Jim were the ones who took an Isis4 I had reversed the wires on in my shop and climbed GFM and replaced it. So, since the input wires were reversed in the canister and GFM is not using S-13s all of the channels are reversed. This was done on 15 November 2016."

The GFM vertical looks reversed. Its more clear in the pql plot when the GFM channel is artificially reversed making it easier to visually match patterns.
Vertical Channels
Vertical Channels with GFM artificially reversed in pql

The GFM north channel also looks reversed. Its more clear in the pql plot when the GFM channel is artificially reversed making it easier to visually match patterns.
North Channels
North Channels with GFM artificially reversed

The GFM east channel is more ambiguous but also looks reversed. Its more clear in the pql plot when the GFM channel is artificially reversed making it easier to visually match patterns.
East Channels
East Channels with GFM artificially reversed