TPrincipal with data sample of different numb. of entries


I would like to use the TPrincipal Class to get a covariance matrix out of my data.
I had a look to the example in the tutorial, but my case is somehow different:
I have a number of correlated variables (say n).
For each variable [size=150]x[/size][size=85]i[/size], with i=1…n, I have one MC data sample with a different number of entries [size=150]m[/size][size=85]i[/size], with i=1…n.
Is it possible to use TPrincipal on such data samples and, if yes, how can I do it?
Or should I have data samples all with the same number of entries (i.e. [size=150]m[/size][size=85]i[/size] = [size=150]m[/size] for each i )?