You could try to make an accurate experimental thermometer by glueing two strips of dissimilar metals together along their lengths. Perhaps try copper and steel or copper and aluminum or steel and aluminum. Try perhaps one half inch wide by six inch long strips. Use sheets of metal that are equally thin. Perhaps cut the strips from aluminum soda cans and steel soup cans, etc. Use a magnet to check for steel vs aluminum. Glue the strips together using epoxy or super glue. Attach one end of the combined strip to a firm base and leave one end free to move in front of a paper scale.
You now have a bimetalic thermometer. The thermometer can be made to be more sensitive by using greater length strips and bending the strips into a spiral and attaching a pointer to the free end of the spiral to amplify movement. This is how most oven thermometers are built except the metals are joined without glue. The principal of operation of a bimetalic thermometer is that one metal expands or contracts more or less than another metal.
When joined, a temperature change causes the combined strip to bend. You can calibrate your thermometer by marking its "pointer" position at two or more large temperature changes then calculating the total deflection divided by the total temperature difference to get deflection per degree (either F or C). Be careful with tin snips and epoxy or super glue. Good luck.
As an alternative, you can make a highly accurate experimental thermometer by winding very thin insulated copper wire (like that used in toy motors) around a wooden spool and connecting the scraped wire ends to an accurate ohmmeter to measure resistance change vs temperature change.
Calibrate as before but remember that calibration temperatures must be steady long enough for any lag to catch up. This is called a resistance thermometer and accuracy will depend on the length and thinness of wire and sensitivity of your ohmmeter. Also the ohmmeter should not cause enough current to flow in the wire to heat the wire. Observe for drift (increase in resistance and temperature) while taking measurements.