Analog inputs all have mysterious positive V offset

[AJB]

For troubleshooting, I'm using the Measurement and Automation Explorer test panel.
When I set the AI properties to RSE and connect ch0 (any channel) to AIGND, the test panel reads a steady 1.247V. Similarly, using NRSE and ch0 connected to AISENSE, the test panel still reads 1.247V.
If I connect a voltage to ch0, the test panel reads that voltage + the 1.247V offset.
I'm using the PCI 6025E, so there are no jumpers to set. The board passed the online diagnostics including self-calibration.

[AJB]

I figured it out. (I'm answering my own question.) One of the load resistors in a current loop was disconnected from the load. With no load, the open-circuit voltage jumped to 24V (supply), delivering 24V to one of the analog input channels. Evidently, if any of my PCI6025E board's AI channels are overvoltage, the rest of the channels will read a constant offset (even if the overvoltage channel isn't part of the sampling sequence).