Densitometers just try to estimate the printing effect, and you hope there is some correlation between the estimate and the result.
In the future if you prints are not coming out right, you can re-check a control strip and see if the color (stain) or gamma has changed from your original control.
Thanks for the advice everyone. Makes sense.
Kirk - quite true about the stain colours. In my case it's Pyrogallol (PMK and WD2D/H). Even then, the stain colours produced by these two Pyro-Metol developers are not the same. PMK tends more to green-yellow while Wimberley's formula is more yellow-orange. Apparently the different alkalis have an impact on stain colour as well.
Reading the negatives the same every time is not good enough imho.
I use a densitometer with the built in settings, you select what you're reading. It's an X-Rite DTP-820, and that has to be calibrated every time before usage.
'Low calibration' (a reading with nothing) which should calibrate to 0.0 0.0 0.0 etc, and a high reading (a set of values usually near 3, you enter the density values of the calibration card, so you can make your own etc).
Even if I come back an hour later, both the high and low readings will have shifted. So if you go back and read the same neg again you will get values that are a bit different.
I always check the calibration. But I must say I have nothing but kudos for the Heiland machine I've been using for several years now. It has proven to be very stable and the reproducibility of readings has been excellent.