Try ASTM E220 "Test method for calibration of thermocouples by comparison techniques".
As well you can try to visit the website of national or international metrologic accreditation bodies. They should have available its own procedures for calibration. I have one for calibration of thermocouples by electric methods, but it's in spanish...
I could send it to you through e-mail if you want.
__________________
It's stupid to discuss about AI: We´ve reached by the "B" way. We' ve producing men as clever as machines.
The best way is to know the the TC you are using. There are many different types. Each one has different ohms measured at freezing and boiling points. This and the types of metals that is used is how they are classified. First find out what type you have then check the resistance in a temperature bath. There are also precision TC's and this will have to be measure in tenths of even hundredths of ohms to determine accuracy. After that it is up to the accuracy of the equipmnet that is used.
Blame it on the validation group. eveyone else does.
__________________
The more complex the mind, the greater the need for the simplicity of play... If you think you can or think you can't you'll always be right
Hi - don't know what experience you have, so this answer assumes you are new to the field. Some of it seems trivial, but you haven't given us much to go on...
First off, beware of 'testing' thermocouples that may be connected to a process that's operating. Disconnecting one that is still working properly can really screw things up. Be aware of the process circuit and it's implications before going ahead. Also, if you have little experience, it's a good idea to get an electrician to ensure that the thermocouple circuit is not connected to some dangerous source of voltage. I came a across a situation where there was over 50V ac on a thermocouple line due to ground faults and poor wiring - bottom line, be careful and consider working with a professional if you really need to learn this yourself. Electrical circuit complacency is one of the worst types of industrial hazard.
If you can disconnect the thermocouple, the first check is continuity. If there is continuity, good. If not - check the wiring leading to the t/c (thermocouple) for breaks. If you can isolate the t/c sensor itself, check it as well.
You should have some idea of what temperature the t/c and process is at. Check a thermocouple table (search for one, or try omega.com) to see what mV you should see for that temperature. Use a good multimeter to check the t/c voltage to see if it matches the table (be sure to use the correct table for the t/c type you are using - eg, K, T, J types...). If you are checking a thermopile, get the appropriate literature for it. As one responder already pointed out, they boost the voltage output level by summing individual t/c elements.
If the t/c seems to put out the correct mV, look at your hardware reading the t/c.
If the t/c reading is wrong, you can swap out the t/c sensor to confirm that the rest of the wiring still reads ok. You can check out your cabling leading to the t/c to ensure that there are no 'pinches' in the wire - if the t/c lead wires do touch at some other point, you will be creating another t/c in parallel with the first one, and effectively averaging the temperatures that they both are seeing.
A quick check on the hardware reading the t/c is to quickly connect a known good t/c to the inputs of the device and bypass the rest of your measurement system. As always, beware of other instrumentation that may be operating at different voltages.
Please be careful, and seek assistance from a professional if needed. Let us know how you make out.