As far as I understand it, I'm *not* changing the physical orientation of the dataset- I'm only changing its file storage format. That's the key part of the problem: I'm trying to get a file format which is easily inter-translatable with ours. As far as I can tell, everything is in the right place- adwarp just fails to either do the expected thing or give an error.
-Joel

The relationship between how data is stored on disk in xyz
matrix format, and how it is sitting in xyz space is, as far
as I can tell, exactly what is meant by the orientation of the
data. The orientaion does not tell you that the right-most
voxel is at 80mm, but it may tell you that the first voxel is
in the right-most plane.

When you say (in AFNI) that a dataset is in RAS orientation,
it means that on the disk, the data matrix has the x-axis (the
fastest changing one) going from right to left, the y-axis
going from anterior to posterior, and the z-axis going from
superior to inferior. That is how the data is stored on disk,
and how it maps to directions in coordinate space.

This is a little different from just talking about where data
is located in space. To know the bounding box of coordinates
(say between 80R and 80L, 83.1A and 79P, and 70I and 90S, for
example) still leaves 48 choices for how it is oriented. The
x-axis can start from any of the 6 faces of the box say, RLAPIS,
leaving 4 choices for y and 2 for z.

So the orientation really implies nothing about the bounding
box, and the bounding box implies nothing about the orientation.
Of course, this is all just terminology. I only hope to explain
what this terminology means in the AFNI world.

- rick