So when you add junctions of differing materials you can create errors. Lets say you add a connector with gold plated contacts. So you go from alumel to gold on one side of the connector. On the other side of the connector you go gold to alumel. The 2 junctions are opposite to each other and generate opposing voltages and assuming they are the same temperature, cancel each other out exactly. In this case they don't introduce an error.
If the junctions are at different temps, you create errors. Let's take the case of a firewall connector where the firewall side of a connector might be 160 degrees F and the cabin side 70 degrees F. There is a temperature difference for the 2 junctions and the voltage generated by the opposing junctions won't be equal and an error will be introduced.
If you splice in copper and never go back to alumel, then you only have one junction and an error is introduced.
If you are running thermocouple wires and want to splice them or run through a connector, it is fine of both sides are the same temperature. If not, you must use splices or connectors made of the same material as the wire.
In most cases we run thermocouple wire from the sensor to the indicator or data acquisition system to avoid unknown temperature gradients in the connections creating errors. Where the thermocouples come off the EGT or CHT, the wires are typically spliced in the engine compartment so both sides of the splices or connections are the same temperature.
When you connect to the sensing system, either an indicator or data acquisition system like the Dynon EMS, use the pins or connection method they call for as they have compensated for this dissimilar junction in their calibration.