An alternative to Mardia-Watson-Wheeler uniform-scores test

#1
Dear all,
this is my first post so please excuse any misunderstanding.
I am working with nonparametric data and I was trying to apply a Mardia-Watson-Wheeler uniform-scores test to 2 or more different classes of data to check if they come from the same population. Unfortunately, some classes have less than 10 elements, therefore the test cannot be applied. Moreover, some of them have just 1 or 2 observations.
I did a rapid search but I couldn't find another test to help me. Does someone know a similar test I can use if I have less than 10 observations per class? I realize 1 or 2 observations is really a small sample and check for statistical significance could be senseless: any suggestion would be highly appreciated.

Regards,
Lisa