Chosing 2 data out of 4

I need some suggestion.
I am working at laboratory that uses mostly ASTM and ISO methods for coal analysis.
For any test sample I make 2 analysis. Then take average result if difference between them is less then repatability limit of the test methods.
Or 2 make two more analysis. Finally I have 4 results.
I am comfused at this point.

Sample data:
Result 1: 0,50
Result 2: 0,30
Method repatability limit: (average)*0,03+0,02=0,032
Result 3: 0,29
Result 4: 0,31

It is obvius that there is problem at result 1. Some corrective action taken.
What statiscal method should I use to choose two data. I think grubbs need minimum seven to eliminate outlier.