Many explanations have been proposed for how these structures formed. The earliest hypothesis was that they were stream channels eroded into the surface of the host rock during the Late Ordovician time when sea level temporarily fell. That explained the presence of crossbedded sandstone... but not the circular shapes or the Presence of Pennsylvanian coal or clay beds.

A second explanation was that they were ancient sinkholes fromed by collapse into a cavern. This idea accounted for the circular shapes and the different types and ages of sediment filling them. Some of the structures may indeed have formed in this way. But in many of the structures, the in-filling rocks are not broken as would be expected if collapse had occurred. In fact, the beds inside some of the structures are horizontal.

In 1950, Missouri geologist J. Harlen Bretz proposed another, perhaps more unusual explanation for how the structures formed, and he named them filled-sink structures, sometimes shortened to filled sinks. In Bretz's model, the structures were indeed the result of groundwater erosion, but without the formation of caves. Rather, microscopic grains of dolomite were dissolved slowly over time, creating tiny spaces in the rock. The loss of rock allowed the dolomite bed to compact under the weight of overlying beds. As it did so, beds above it sank very slowly into the structure. The amount that younger beds settled varied, depending on the amount of dolomite rock mass removed: the more rock removed, the greater the subsidence... Once uplift of the Ozark Dome began in earnest in Late Pennsylvanian time, the water table across the Salem Plateau dropped, ending the process...

Bretz's model accounts for the observed characteristics of many of the structures...

Be that as it may, some circular features found in Ordovician dolomite in the Ozarks are indeed attributed to collapse of deeper caves. These structures, called "circles," are similar in shape to filled sinks but always contain broken, tilted, and shattered beds rather than horizontal strata.