Assigning outliers in very few datapoints

#1
Hi,

I'm a geology PhD student and i have a problem with making my dataset statistically robust.

I'm measuring element concentration of my samples using XRF. I have about 900 samples and measured each sample 3 times. Of course there is a variance in these 3 measurements belonging to each sample, and sometimes it's clear that one measurement is way off (say 50% ) compared to the other two. However I am trying to find a statistically robust method to find those kind of outliers for each set of 3 measurements.

I tried to find a method on the internet, but all examples only consider very large datasets. I only have 3 datapoints so calculating the mean or quartiles seems futile to me. Also, since the variance between samples is much bigger compared to variance within one sample, i cannot use the whole dataset, but have to look at each individual sample.

The questions i have are:
Is it possible to find outliers in such small datasets? if so, how?
How can i calculate my cutoff (i.e. above which i should consider a datapoint as an outlier)?
What is the most conventional way to deal with these kind of problems and do you have references so i can make my dataset suitable for publication?

I would like to thank you already for your time and effort to help me out.
Kind regards,
Bas