It seems to me that we took the strain rate to the other side, and then since we cannot really "divide" by such a complex quantity, we do something like a "multiply and divide" by S(u).