I just found this on hackaday recently, it sounds fascinating how you can use a quartz crystal, which when a weight is applied changes the frequency enough you can get extremely precise readings.
One thing I'm wondering though, is aren't quartz crystals effected by temperature a lot? Is there a simple way this would be compensated?
I know with some radio devices (I'm specifically thinking of RTLSDR hacks) a heating element is added to maintain a constant temperature (or a TXCO -- a temperature controlled crystal oscillator).
It depends on the way the crystal is cut, usually they are optimized to have a maximum around room temperature so that effects from variations are minimized. The effects might partly be compensated by measuring the temperature and applying a correction to the data since the temperature dependency is known, but the sensitivity is also affected. So, temperature control is definitely important for accurate measurements.