I've made some changes to KPS1_CLPA2 that should fix this. On 23/07/07, Malcolm J. Currie <[log in to unmask]> wrote: > > the problem seems to be coming from KPS1_CLPA2, which does some further > > processing of the NDF WCS and then re-writes it in a broken form. Baffles > > me what the purpose of this routine is, but it should just leave this WCS > > alone. Au contraire! Without it, collapsing a 3D cube along an ra (or dec) axis results in a 2D pixel array that only has a single (spectral) WCS axis. It's purpose is to introduce extra WCS axes (which are just copies of the corresponding pixel axes) so that collapsing a 3D cube always results in a 2D pixel array that has 2 WCS axes. David