Hi, Concatenating the wrong way around only makes a difference when the 3D transformations are not co-planar. If most of yours were nearly co-planar (with little shift/rotation in/about y) then there wouldn't be much difference between the two ways of doing it. So this is a normal, expected result. Glad it works now. All the best, Mark On 25 Jan 2007, at 23:13, Joerg Magerkurth wrote: > THX Mark! > > I have evaluate about 40 datasets with this mistake in the scripts. > But only > at two datasets it was visible! Now it works with all. The whole > procedures > are scripted with perl. I have changed this line by now after your > mail. Do > you have an idea why the wrong script worked with so many datasets? > The dataset were from two studies. It's crazy! > > Thanks for help! I hope I can pay back by time. > > Best > > Joerg