How to compare two PCA vectors - classification and clustering problem

Dear all,

Assume I have one reference PCA vector and two sample PCA vectors, all of the same length. I want to know which of the two sample PCA's is more similar to the reference vector.

Do I just subtract one sample from the reference and the other sample from the reference and compare the results? This seems to ignore the fact that the values in each vector have decreasing contribution to the overall variance...

Kind regards,