For more than a century geologists have wondered why some bedforms are orientated roughly transverse to flow, whereas others are parallel or oblique to flow. This problem of bedform alignment was studied experimentally using subaqueous dunes on a 3–6-m-diameter sand-covered turntable on the floor of a 4-m-wide flume.

In each experiment, two flow directions (relative to the bed) were produced by alternating the turntable between two orientations. The turntable was held in each orientation for a short time relative to the reconstitution time of the bedforms; the resulting bedforms were in equilibrium with the time-averaged conditions of the bimodal flows. Dune alignment was studied for five divergence angles (the angle between the two flow directions): 45°, 67–5°, 90°, 112–5° and 135°. The flow depth during all experiments was approximately 30 cm; mean velocity was approximately 50 cm s^-1 and mean grain diameter was 0–6 mm. Each experiment continued for 30–75 min, during which time the flume flow was steady and the turntable position changed every 2 min. At the end of each experiment, water was slowly drained from the flume and dune alignment was measured. Transverse dunes (defined relative to the resultant transport direction) were created when the divergence angle was 45° and 67–5°, and longitudinal dunes were created when the divergence angle was 135°. At intermediate divergence angles, dunes with both orientations were produced, but transverse dunes were dominant at 90°, and longitudinal dunes were dominant at 112–5°.

One experiment was conducted with a divergence angle of 135° and with unequal amounts of transport in the two flow directions. This was achieved by changing the orientation of the turntable at unequal time intervals, thereby causing the amount of transport to be unequal in the two directions. The dunes formed during this experiment were oblique to the resultant transport direction.

These experimental dunes follow the same rule of alignment as wind ripples studied in previous turntable experiments. In both sets of experiments, the bedforms developed with the orientation having the maximum gross bedform-normal transport (the orientation at which the sum of the bedform-normal components of the two transport vectors reaches its maximum value). In other words, the bedforms develop with an orientation that is as transverse as possible to the two flows. In those cases where the two flows diverge by more than 90° and transport equal amounts of sand, bedforms that are as transverse as possible to the two separate flows will be parallel to the resultant of the two flow vectors. Although such bedforms have been defined by previous work as longitudinal bedforms, they are intrinsically the same kind of bedform as transverse bedforms.