"That's probably why there are so lamentably few replication studies done," says Bella. "It's a real problem, actually - the people qualified to redo studies aren't interested, so they don't, so we go by old results that one study got one time, and for standard p-values that means that one in twenty results we're relying on is just wrong. Coincidence." She looks at their lab experiment with a small sigh. "One thing I can say for this kind of silly make-work is that if it turned out these dialysis membranes didn't consistently work as advertised? Some undergrad class would've noticed by now."