Jets of iron

Did you ever, when washing dishes, shove a funnel wide-side-down into the water to see how high the water would jet up from the narrow end? Or, in the tub, put your palms together underwater and quickly squeeze them tightly together, to see how far the water would shoot? When you're dealing the continent-sized chunks of molten and solid iron moving around, it seems you can get fast-moving jets of iron under the mantle. Of course "fast" is a relative term: O(40)km/year isn't going to rival the jet stream, but it represents quite rapid progress through rock.

The image at the site is a little misleading--it's just a toy model of what things might be like. Nobody really thinks there are cylinders like that in the core.

The discovery of the jet involved tracking two massive but unusually strong lobes of magnetic flux originating from the core-mantle boundary, situated beneath Canada and Siberia respectively, but moving with the flow of the molten iron. Because their motion could originate only from the physical movement of molten iron, the lobes served as markers, allowing the researchers to track the flow of iron.

Livermore likens it to being able to track the course of a river at night by watching candles floating on the surface. “As the iron moves, it drags the magnetic field with it,” he says. “We can’t see the flow of iron itself, only the motion of the flux lobes.”

Three satellites sensing the magnetic fields as they orbit the Earth found some variations with time, and it looks like these variations come from the mantle-core boundary.

I don't know if seismography would have the resolution to confirm this.