One needs in talking about thermometers understand the difference between
Most digital thermometers offer very high refinement in readability and modern ones a relatively good repeatability for readings of 0.2C or finer. They are, however, typically inacurate and most use simple chipsets and assume a circuit and probe linearity that they don't really have. One can, in praxis, calibrate them to acurately read out at one or two temperatures. That's where a good reliable, repeatable and traceable reference comes into the picture.
The Kodak process thermometer type-III is NIST traceable and has quite a few calibration points in both C and F. They cover a larger range, are made of heavy glass and protected by a nirosta sheath so they are very hard to read but very acurate at the known points. They offer moderate repeatability due to their poor readability. Their advantage is the large range, robustness and economical price---- just a few calibration points against a NIST traceable reference will cost much more.
I also have several Jobo photographic thermometers they call "color". They have good readability and repeatability but very poor acuracy. My mercury Jobo is a bit better than the spirit one but the spirit version is much easier to read. They are OK for uncritical B&W processes but not all B&W processes are uncritical. Once I started doing some process control I had to realize that these thermometers were often insufficient.
I also have a set of high precision laboratory thermometers for use in the 20C range and readable to resp. 0.1 and 0.2C. They too are, of course, mercury and very very thin--- and are stored in special cases. Since these thermometers are very acurate and I'm really mainly interested in only a few points I use them to calibrate my other thermometers both spirit and digitial--- including the digital temp built into my Nova 2000 controller--- in that range and use one of my Kodak type-IIIs to get some other points, especially the "magic" 100F. Calibrated against these standards the "inexpensive" digitals are quite adaquate and appropriate to the task and suited to even the most stringent of process control regimes.
From Google enter, vee gee 81125 . Or go www.novatech-usa.com.
I may buy a couple of 1 or 1.75 inch Vee Gees; 25 to 125 F.
Quick, dial/stem thermometers; $4 and $7.
I droped the one I had. Dan