The reason many of our solving techniques such as Skyscrapers, 2-string Kites and other chaining processes work, is because they examine situations where there are only two possible options, and both lead to the same result somewhere else in the puzzle. So we can take that result as true, even if in the meantime we don't yet know the true arrangement that predicts it. I often us this approach, which I call SEW. Try this:

Here I was looking at what was going on in Box2. With the 4,5 Locked Pair on the middle shelf, there were only two places for the 6. They worked out as

Now what does that mean for the rest of the puzzle? It turned out that the first scenario predicted Row5 to be 297 641 358 and the second forced it to be 297 346 158.

This meant that in either case the 6 in Box5 had to be on the middle shelf, which in turn meant that the 6 in Box6 had to be at r4c8 anyway, which easily solved the whole puzzle.

There are of course many ways to solve this puzzle and you can look up the discussion of it which took place between ravel and RW in 2005 in the post Solving without pencilmarks.