Well that link is quite read. OK, I think I get what you are saying there. I did initially put all ballnuts in the middle of the axis, at the centre of stiffness. I hadn't though about it in the way the article suggests, it was just an intuitive start point, but the flange on the ballnut means they need alot more space which pushes everything apart.

So the X axis ballnuts are now on the 'front' of the gantry, whereas the article suggest they should be further back in the middle of the gantry. Problem is that they would hit the X bearings and make everything wider which knocks onto the frame and bed. I've only got so much space and I had wondered about an enclosure in the future so don't want to go past the edge of the table with any parts.

The Y axis ballnuts are also on the side plate of the Y axis, not in the middle of the bearings. Again the flange would push them higher in Z, and they needed much more material under the bearing mounts at each end of the ballnut to support them. It all looked too tall when I drew it.

To summarise I think what the article is saying is that because things will twist and rotate, you should put the ballnut in the centre of rotation (that is rotation of the structural parts caused by moments) so that the nut sees minimum binding loads etc. I think that is a good principle to aim for where ever possible, however I think in practice the machine I've drawn will see relatively small loads at the ballscrew due to this deflection causing a radial / binding load on the ballnut and I'm not too concerned.