Why is it so difficult to catch a fly? The amazing, and often irritating, aspect of an insect’s flight is not so much the animal’s speed as its manoeuvrability. Aeronautical engineers would love to know how the fly manages to be so evasive. Now the mystery seems to have been solved, thanks to ‘Robofly’, a scaled-up model of a fly in motion.

This mechanical insect, made by Michael Dickinson of the University of California at Berkeley and collaborators, is not a complete scale model of the full creature, but a pair of wings powered by motors. The device is programmed to mimic the beating pattern of flies’ wings, and the researchers are able to measure the lift forces that this generates using force sensors situated where the wing joins the ‘body’.

Because air flows differently around a large mass than around a small one, the team had to use a different, more viscous fluid to mimic the airflow pattern around their scaled-up wings. So they immersed the wings in two tons of mineral oil.

What they found, as Dickinson told a meeting on ‘Bionics and Biomimetics’ at the Institute for Advanced Study in Berlin, Germany last week, was that conventional ideas about insect flight aren’t enough to explain how the fly stays airborne.

In 1996, Charles Ellington and coworkers at the University of Cambridge used their own version of Robofly - a giant mechanical moth - to answer the long-standing puzzle of how insects stay up at all. In a report in Nature, Ellington and colleagues showed that the secret is in the flapping. Everyone knew, of course, that insects flap their wings while jumbo jets don’t, but no one could understand what effect this had. The Cambridge mechanical moth revealed circulating vortices of air moving along the upper wing from the body to the tip, like little tornadoes turned sideways, during the wings’ downstroke. These vortices push air downwards behind the trailing wing edge, and so generate lift.

The vortices are produced through a cunning gamble in which the insect places the wing briefly at a steep tilt to the airflow. If prolonged, this makes the flight stall. But if applied only briefly, it generates the helical airflow pattern before stalling kicks in. So this mechanism of flight is known as ‘delayed stall’.

But, says Dickinson, delayed stall can’t be the whole story. For one thing, it generates only just enough lift to keep the insect aloft, whereas in reality flies can be loaded with twice their body weight and still fly. In addition, delayed stall doesn’t explain why flies are so maneuverable.

And to drive this point home, Dickinson’s Robofly showed that the lift forces during flapping show sharp increases during the turning points from downstroke to upstroke (and vice versa) which delayed stall can’t account for. Somehow, the fly is augmenting its lift in other ways.

Dickinson and colleagues have identified two additional mechanisms for creating lift during the wing stroke. First, the wing rotates as it passes from upwards to downwards motion. (This produces a lift force in just the same way that a cricket ball with backspin will wobble upwards when a bowler delivers a ‘googly’.)

The second new trick for creating lift is even more striking. As the wing rotates, an air vortex comes spiralling off from the edge, just as a slowly moving boat in still water can create vortices in its wake. These vortices are like little packets of energy, passed from the wing-driving mechanism into the circulating fluid. But the fly’s carefully timed beating pattern allows it to recapture some of this energy from the vortex before it spirals out of reach. Dickinson and colleagues call this ‘wake capture’. They describe their results in the 18 June issue of Science.

Rotational lift and wake capture give the insect a lot of control over its flight. By making only small changes in the timing of wing rotation during the beating cycle, the fly can produce substantial changes in lift - even generating negative lift (a downwards force) if needed for manoeuvring. While most of the studies of Dickinson and colleagues have emulated the flight dynamics of the fruit fly, they have also shown that very short strokes, like those of the hoverfly, make the two new lift mechanisms the main contributors to the lift force: delayed stall hardly counts at all. This, then, would explain why the hoverfly is able to dart around in the air with such stunning precision.