Callibration problems

These thermometers were all supposed to be showing the same temperature. How would we know which one is correct? There is a range of almost 2oC. Each one might be correct to the nearest 0.1oC but I'd suggest that an uncertainty of +/- 1oC might be more appropriate based on what are effectively repeat readings. We'd need to put them in melting ice and into boiling water to check them properly.