Hi Richard,
I agree with you in terms of collecting a lot of historical data to try to make some sense of the the latest data.
From what I have heard, I think this test has far too many variables. (Water quality as you advised, tramp oil on the surface of the sample, sample surface vibration, sample temperature, the condition of the thread, air currents at the surface of the sample, repeatability each time the thread is dropped, plus of course pH, lubricant concentration and so on.)
Cheers
Peter J. Stewart-Hay
Principal
Stewart-Hay Associates
www.Stewart-Hay.com
519 641-3212