Why not to use emissivity tables!
During training we often tell people that they should measure the emissivity of a material if they need to accurately measure its temperature, why is this?
There are many different emissivity tables available from various sources including universities and camera manufacturers. If you look closely at them you might notice different published values for the same material. This happens for various reasons. Firstly, and it is obvious, the sample used for testing might differ. Secondly, the camera spectral response may be different between the cameras used. Other factors like differing object temperature and environmental factors will also be of influence. A lot of people think that the result will be the same with a single sample from longwave camera to longwave camera, and this is not true. The exact spectral response of the camera might be different, and while it is true that a bolometer has a fairly liner response, the lens coatings will differ, and they may operate in slightly different parts of the spectrum. There might not be much of a difference for some materials, but for others it may be very significant.
Today a lot of cameras have an onboard emissivity table, but it is not the case that the materials listed were all measured with that camera. In fact it is generally a standard emissivity table just programmed into the camera.
Problems occur when thermographers believe that the values in tables are absolute values and can be universally applied. This is simply not the case. There are also problems with the interpretation of the material itself. What one person might describe as shiny, someone else might describe as polished etc.
In a classroom environment we often get students to measure the emissivity of sample materials, and the measured values can vary significantly. This is sometimes due to the difference between the detectors in the cameras, or the temperature of the object might be different, or maybe even they take their measurement from a different location on the sample. This happens when they compile tables also, and they usually publish an averaged value.