Hello mwbm07,
If there's no information in the Reference pane and therefore no [R] or [T] mark next to the chunk's label in the Reference pane, then the chunk will remain unreferenced and all the coordinates and dimensions of the produced results would be in some arbitrary internal coordinate system.
In case the chunk is referenced in local or geographic coordinates the additional 4x4 transformation matrix (chunk.transform) will be introduced that performs the transformation from internal coordinate system space to geocentric (in case of geographic referencing) system or to local coordinates if it has been selected for referencing.
Via Python you can access the internal coordinates of camera locations, marker positions, coordinate of the mesh polygons, tie point locations and etc.:
camera.center, marker.position, for example.
You can check the following topic that has an example of setting the coordinate system of the chunk (origin and orientation) according to the location and orientation of the selected camera from the aligned set:
https://www.agisoft.com/forum/index.php?topic=11120.0