I found the schematic below for a muxed thermocouple circuit from this web page. This circuit uses a mux to switch just one leg of the thermocouples. While I am sure that this works fine I have a feeling that its causing a couple problems with the noise immunity that one would normally get with a differential signal. First, each leg now has a different impedance to ground which could cause different voltages to be induced on each cable when interference is present. Second, all cabling of the disconnected thermocouples will behave as an antennas inducing noise on the negative input for all of the other channels.
Is this a correct assessment? And, besides using a differential mux, is there a reasonable low complexity, low component count way to accomplish this (muxing thermocouple channels)?