When using rpy2 2.1.9 and numpy 1.5.1, the attached script segfaults. GDB trace attached.
The operating system is 64-bit Ubuntu 11.04.
Although the segfault is a bug, pca_loading is an R list which cannot be exposed as a numpy array.
What I find a bit weird is that replacing the line
pca_loadings = r.princomp(variates).rx("loadings")
pca_loadings = dict(r.princomp(variates).iteritems())["loadings"]
makes the numpy conversion work without a segmentation fault.
pca_loadings = r.princomp(variates).rx2("loadings")
Nothing weird to me: the wrong kind of input what not handled properly and make the code crash, the right kind of input is working fine. Consider using the mailing-list, for example, if questions.