Heyhey,, Good exploitation of the rather peculiar UV mapping of Roxie! 

Not sure about the usefulness of having the tongue, teeth and eyes in the same map, but the advantage of having head and body in the same texture map is obvious.

Allow me to cheat the magic:

Roxie's body is mapped in the (1.0, 0.0) (2.0, 1.0) region in the UV space. Poser wants to see co-ordinates in the range 0.0-1.0 and uses the modulus if the u or v co-ordinate is outside that range, so a body texture vertex (1.345, 0.678) in the .obj is translated to (0.345, 0.678) to look up the rgb values so in the standard setup Poser finds what it must find.

When applying Vilter's trick Poser considers the wide texture of course as the (0.0,0.0) to (1.0, 1.0)  UV space. The scale factor in u of 2.0 makes Poser look for rgb value (u,v) at (0.5*u, v), and so, again, find what it must find. 

In principle same trick could be played in the V-direction, having multiple skins in one texture map.

Again, Bravo! for the invention!