Hi,
As with every measurement device:
* There is a theoretical optimum
* and there are errors
Then length of a wire - per se - is not the problem. But introduced noise, impedance, material, leakage current ...
And amplifiers also introduce errors (offset, gain, linearity..), and the cold junction compensation, the software, the linearisation....
Nothing special to thermocouple.
If you want high accuracy, the effort becomes huge.
***
All the physics and math is known. But some designers simply don´t care that much about it.
Some killers for thermocouple performance:
* wrong cable (they need to have the same two materials as the thermocouple itself)
* noise, and/or missing filters
* (offset) drift of OPAMP
* wrong software/linearisation.
Klaus