I found that dripping water from an eye dropper into each side to compare was quite unreliable/variable - ie results did not have good repeatability (part of which is probably cause by losing count - any distractions and I was lost!). I then calibrated it using the drip method recommended by Davis using 43ml of water. However, that only represents 2mm of rainfall, and when each bucket holds 0.2mm - that’s an error of 10%.
I am now comparing it to a manual gauge mounted right up against the Davis during good rainfall (and it’s lucky we’ve had a bit of that recently). I have checked the calibration of the manual gauge, and it seems good (Davis Australia were not happy that their gauge was being checked by a mass-produced item :D). I am gradually winding the adjustment screws up so my Davis readings are scattered each side of the manual gauge reading (rather than all on the “less” side).