Accuracy of thermometers?

I recently tried an experiment of measuring the temperature of some hot water with several different thermometers to see which one was most accurate. I used 2 dial thermometers and 2 digital thermometers-a TempRite 35189 and a Comark 300 BSF (neither have a calibration). I find it is hard to get a similar reading among the thermometers. The dial thermometers seem to be a bit more on par with each other while the digital ones seem to sync more. Can anyone give me advice how to get the most accurate readings out of a thermometer. I really want to be as accurate as possible with my mash temps as well as sparge and strike temps.

Dial for me... I spent four years working in a semi conductor laser lab where temperature variations played a huge role during our prototyping phases. We always used dial therms... even when integrating our power variations through labview software.

Not to say digital isn't equal or better in some cases... Ive just personally seen more consistency among "mechanical" apparatuses, than digital when measuring LARGER ranges of temps... lab geek over here.