Did a 'proper' (although single) test. Started with a pan full of iced water and checked the glass thermo read zero C, then removed the ice and gently heated the pan until boiling and took a reading of the digital every 5 degrees according to the glass thermo. Then turned off the heat and took readings every 5 on the way down (but stopped after about 70 as the readings were almost exactly the same as the way up, which is good as it shows there isn't much lag).
Conclusion ? For this particular thermometer the size of the error depends on how close the temperature is to 45 and whether it is above or below 45. Below 45 it reads high, above 45 it reads low. It's roughly 3 times more inaccurate above 45 as below and the error is linear (straight line) above and below with the error increasing as it gets further from the magic 45. At zero it read 1.8 at boiling it read 92.8.
So what does this mean for using it ? It's really handy to have a device like this with its relatively large display so you can monitor the temperature of a liquid in a container (HLT, mash, FV) just by glancing across at it. In practical terms, though, I wouldn't want to use it to gauge things like strike and sparge temperatures without knowing its error. For instance with this unit 73 is shown as 69 and 82 is shown as 77. So I'm going print out a graph and stick it on the wall so I can quickly check the true temperature.
Of course I'm now wishing I'd had all three thermometer probes in at the same time, but I suspect the errors will be similar across all three.
:geek::geek::geek: