How about "the ice caps are made of salt water"? In that case it was not semantics, they were trying to teach us that the polar ice caps were made saltier water than the surrounding oceans.
And if they use an experiment to demonstrate something other than what the experiment actually shows, is there a point in wasting time and effort on the experiment? Such as using a balloon inflating on the reaction between baking sofa and vinegar to demonstrate an imperfect system when the primary factor in decreasing mass is the increasing volume which increases buoyancy. And if you calculate the mass of the air that would be in the increased volume, you get the approximately discrepancy. (Imperfect measurements yield imperfect results.)