The rearrangement number is a quantity in set theory that measures how many reorderings of the natural numbers you need in order to "disrupt" every conditionally convergent series. A conditionally convergent series is one that adds up to a finite value, but would diverge if you took the absolute values of all its terms. A classical result in analysis says that by reordering the terms of such a series, you can make it sum to anything you want. The rearrangement number asks: how large does a collection of reorderings need to be so that, no matter which conditionally convergent series someone hands you, at least one reordering in your collection changes its sum or makes it diverge?
Set theorists care about comparing such quantities, called cardinal characteristics, to each other. Previous work established that the rearrangement number sits between two other well-studied cardinal characteristics, one related to measure theory (covering the null sets) combined with another called the bounding number on one side, and one related to category theory (the non-meager number) on the other side. The natural question was whether these inequalities can be strict, meaning whether the rearrangement number can genuinely differ from the quantities bounding it, or whether it is always forced to equal one of them.
The authors answer this open question by constructing two different mathematical universes, using a technique called forcing, in which the rearrangement number is strictly separated from the bounds in each direction. In one universe, the rearrangement number is strictly smaller than the non-meager number, and in another it is strictly larger than the combination of the other two characteristics. These results show that the rearrangement number is a genuinely independent and flexible cardinal characteristic, not reducible to the previously known ones. Along the way, the authors also pin down related consequences for two other cardinal characteristics, the subseries number and the splitting number.