Do you know, please, why is package "hier.part" limited to the maximum number of 12 variables? When is hierarchical partitioning used for large-scale comparisons, I found this limitation very un- usefull. I tried to somehow split my data to less variables and computed the hierarchical partitioning on limited data, but the results were more or less different, especially in joint effects. Then I changed in the functions "all.regs", "partition" and "hier.part" the amount of used variables. My 15 variables needed some 60 thousands of regressions and it needed some time (in my machine about one hour of work, but it is not bad), but after this the program terminated with some error message. I do not have any idea, where could be the problem. One problem occured in calling function "all.regs", where were problems with allocation of dynamic memmory, when using "print.vars = TRUE". But it again terminated calling the function "partition". Do you know, how to solve the problem and, additionally, is it possible to do the regressions more heuristic? With the best regards Zdenek Fric