Getting a bit off-topic, but still:
the relative strength of back and belly (including sinew backs) tremendously depends on relative humidity, and on how different woods react in tension and compression forces to such changes.
Note the below graph from the wood handbook:
A: tension strength of wood peaks at 12% MC, and decreases at lower MC and higher MC
C: Compression strength of wood increases with decreasing MC.
The 1/3 2/3 rule of thumb (rule of thumb, not the absolute truth) holds for regular 12% MC, no heat-treated bellies. Some woods are stronger in compression, there it will be closer to 40-60 at 12% MC, and it may shift to 50-50 at lower MC, and even 60-40 at very low MC.
That's why bows gain poundage at lower MC, and break at very low MC: the belly overpowers the back.
Most of the time, the back overpowers the belly, and we get set. A bow with set is still a bow. In reverse, a belly can only overpower the back once...
Sinew, however, will behave in tension like wood in compression: its MOE (stiffness) decreases with MC.
Sinewing a bow is like trapping the back, but in a safe way: you make the back surface weaker, hence the belly is working less. In a trapped back, the neutral plane moves towards the belly, relative to an untrapped back. It's safe because it's darn difficult to break sinew. If, however, the sinew layer is too thin, you'll still break the back when the wood fibers underneath the sinew are overstretched.
If you sinew an existing bow, the neutral plane will shift a bit towards the back. But not as much as it would for a same draw weight all-wood bow to which you would add a wood backing. The reason is that nowpart of the draw weight comes from overcoming the shrinkage of the sinew which put the belly under tension. This is similar in theory to a one-sided perry-reflex. (not that this clarified anything
)
If you really want to know how things work the way they do, walk the walk: perform predictions on how something's supposed to work, make bows (or test samples) to test these predictions (not one of each kind, but at least three), and see if the predictions worked out.