In Minnesota’s Judge C. R. Magney State Park, you will find one of nature’s mysteries known as Devil’s Kettle. As the Brule River makes its way through the park en route to Lake Superior, it is bifurcated by a large rock formation creating a pair of waterfalls. Nothing too out of the ordinary, expect for the fact that nobody can figure out where one of them goes. See, the first fall continues on to the lake but the second dumps into a large pothole and vanishes.
Geologists have studied the formation and cannot figure out where the water terminates. Not only do they not know where, they can’t even come up with an agreed up explanation of how. John C. Green explains the problem in his book Geology on Display:
One [theory] is that, after dropping down the pothole, the river runs along a fault underground, or as a variant, that it enters an underground channel and comes out somewhere under Lake Superior. Both of these ideas have one valid aspect in common: they recognize that water must move downhill! But the main problem is creating a channel or conduit large enough to conduct the impressive flow of half the Brule River! Faulting commonly has the effect of crushing and fracturing the rock along the fault plane. This could certainly increase the permeability of the rock — its capacity to transmit water — but the connected open spaces needed to drain half the river would be essentially impossible, especially for such a distance. Furthermore, there is no geologic evidence for such a fault at the Devil's Kettle. Large, continuous openings generally do not occur in rocks, except for caves in limestone terranes. The nearest limestone is probably in southeastern Minnesota, so that doesn't help... Maybe the Devil's Kettle bottoms out fortuitously in a great lava tube that conducts the water to the Lake... Unfortunately for this idea, they are not the right kind of volcanic rocks! Rhyolites, such as the great flow at this locality, never form lava tubes, which only develop in fluid basaltic lava. Even the basalts in this area may not be the "right kind", being flood basalts that spread laterally as a sheet from fissures, not down the slopes of a volcano. No lava tubes have been found in the hundreds of basalt flows exposed along the North Shore. Furthermore, the nearest basalt is so far below the river bed, and even if it did contain an empty lava tube (very unlikely after its long history of deep burial) the tube would have to be both oriented in the right direction (south) and blocked above this site so that it isn't already full of debris. And there are no reports of trees or other floating debris suddenly appearing at one spot offshore in Lake Superior. The mystery persists.