It is important to make sure the temperature readings are correct in your oven/furnaces before making any interpretation of the readings. Therefore, if some furnaces have not been used for a long time, it is necessary to test the thermocouple readings before starting any new experiments.
Melting materials with well-known melting points is a good way to check the thermocouples. We recently used Au (1064 C at 1 atm), LiSiO3 (LSM, 1203C at 1atm), and diopside (MgCaSi2O6, 1391C at 1atm) in the lab to test the TC over a wide range of T of a 1atm furnace in the lab. The basic setup is to hang these materials either in a Pt wire loop or a Pt basket in the furnace and observe at what reading temperature they melt.
For Au, it is convenient to make a piece of gold “flag” and have it on a Pt loop. When Au melts, the flag shrinks into a little ball, which can be directly observed from the bottom window of the furnace if a mirror is well positioned below the furnace.
Au melting observed from the bottom window of our 1atm furnace at LDEO experimental petrology lab.
We put LSM or diopside in a Pt basket and constantly take them out of the furnace to see if melting happens when incrementally heating (1-2 degrees each time) in the neighborhood of the melting temperature. LSM does not quench as well as diopside in this procedure. So we started LSM as crystals and carefully increased T until we see a puddle of quenched crystals instead of a bunch of individual crystals. Diopside quenches pretty well, so it is quite easy to distinguish if it crystallized (powder_ish white) or melted (very clear glass). Both LSM and Au melting happen within quite small T intervals (1-2 degrees) when the melting process of diopside is usually observed within (2-4 degrees).
This is a pretty simple procedure, but we did learn some small points along the way that we hope we had paid attention to from the beginning. First, make sure the TC is actually positioned at the hotspot of the furnace. If it was not, the reading temperature would be very different from the setpoint temperature. Second, make sure the sample is hanging at the same level where your TC is positioned. Hanging well below the TC will falsely increase the T correction because the sample is actually sitting somewhere colder than the reading temperature. Last, depending on the amounts of materials, wait for at least 15-25 minutes for each temperature interval before making the melt/no melt observation.