> and see what the input impedances are...
In 1928, when tubes were expensive and gain was hard to get, most boxes used input transformers to step-up the voltage and provide "free gain". And for maximum flexibility in patching, including patching through long (telephone) lines, everything was 500 ohms, later 600 ohms except ABC used 150 ohms.
Today with $0.10 chips (actually since the 1940s), we are not hurting for gain and "most" gear has "high impedance" inputs. Line inputs are usually 10K to 50K. We still tend to have outputs able to drive 2K down to 600 ohms, so we can simply Y-connect many 10K inputs to one output.
Yet there are still a few old-time 600 ohm inputs. Passive EQ are often true 600. Fairchild 660 is true 150/600. A modern studio won't have old junk like this hanging around, but guys who play with AM17s might.
Langevin was one of the last of the old-style audio systems. Gear like the AM16/17 may have 150/600 ohm inputs, and was intended to have a full load, often a 600 ohm attenuator. It will "work" with a 10K load, and the midband gain is about the same, but Joe says (he's surely right) that small errors in the transformer, tuned to work perfect with a 600 ohm load, will ring-out the top of the audio band when the load is much greater than 600 ohms.
On these amps, 500 or 1,000 ohms is "loaded". 10K is "barely loaded" and will be "ringy" as Joe says. As Scott says, sometimes the "sizzle" from "mis-loading" is musically useful. In a complex busy studio, sanity suggests putting a 650 ohm resistor on the output permanently, and only using it with hi-Z inputs. You can always get a top-end ring with an EQ. Situations where a producer "must have" the AM17 unloaded-ring, or must interface a true 600 input, can be handled as special cases with a little extra money for the electronics technician.