Bonjour,
Variation of your result might be caused by different factors (work technique, environnement, instrument). However, from my point of view, the most porbable source of error is sample preparation.
Are your samples well shaken pior to analysis? Are the air bubbles taken out after shaking? How much time separates shaking from sampling?
As you know, water (particulary at high concentration) will not be soluble or miscible in the oil. This means that you`ll have heterogeneous solution (water bubbles will be suspended in the fluid). Thus, it is needed to be particulary cautious when preparing the sample for analysis.
As for your sample 1, there is so much water in it than even your lowest water concentration (1.8%) raises concerns and requires immediate actions (filtration, oil replacement, identification of water source). Do you really need more precise results in this case? High water content in this sample would explain result variation.
Sample 2 results are better (test 1 and 3 are pretty close). You could either do a 4th test, discard test 2 or make an average.
This will depend on the sensibility required, good laboratory practice and common sens.
I hope this helps,
JFG
