Interesting piece of hardware! Some Googling suggests that one was used in a
quantum magnetometer (an experimental one, at that) back in the mid-1960s. That might explain why it doesn't have a traditional lock-in loop.
I don't see any references to the model number in
Bill Riley's book, but he does cite an
early-60s Varian patent for a device that looks a lot like it. This patent does describe a complete Rb frequency standard. No schematics, unfortunately.