Morphological parameters are presented for a variety of insects that have been filmed in free flight. The nature of the parameters is such that they can be divided into two distinct groups: gross parameters and shape parameters. The gross parameters provide a very crude, first-order description of the morphology of a flying animal: its mass, body length, wing length, wing area and wing mass. Another gross parameter of the wings is their virtual mass, or added mass, which is the mass of air accelerated and decelerated together with the wing at either end of the wingbeat. The wing motion during these accelerations is almost perpendicular to the wing surface, and the virtual mass is approximately given by the mass of air contained in an imaginary cylinder around the wing with the chord as its diameter. The virtual mass ranges from 0.3 to 1.3 times the actual wing mass, indicating that the total mass accelerated by the flight muscles can be more than twice the wing mass itself. Over the limited size range of insects in this study, the interspecific variation of non-dimensional forms of the gross parameters is much greater than any systematic allometric variation, and no interspecific correlations can be found. The new shape parameters provide quite a surprise, however: intraspecific coefficients of variation are very low, often only 1%, and interspecific allometric relations are extremely strong. Mechanical aspects of flight depend not only on the magnitude of gross morphological quantities, but also on their distributions. Non-dimensional radii are derived from the non-dimensional moments of the distributions; for example, the first radius of wing mass about the wing base gives the position of the centre of mass, and the second radius corresponds to the radius of gyration. The radii are called `shape parameters' since they are functions only of the normalized shape of the distributions, and they provide a second-order description of the animal morphology. The various radii of wing area are strongly correlated, as are those of wing mass and of virtual mass: the higher radii for each quantity can all be expressed by allometric functions of the first radius. The overall shape of the distribution of a quantity can therefore be characterized by a single parameter, the position of the centroid of that quantity. The strong relations between the radii of wing area, mass and virtual mass hold for a diverse collection of insects, birds and bats. Thus flying animals adhere to `laws of shape' regardless of biological differences. Aerodynamic and mechanical considerations are most likely to provide an understanding of these laws of shape, but an explanation has proved elusive so far. The detailed shape of a distribution can be reconstructed from the shape parameters by matching the moments of the observed distribution to those of a suitable analytical function. A Beta distribution is compared with the distribution of wing area, i.e. the shape of the wing, and a very good fit is found. With use of the laws of shape relating the higher radii to the first radius, the Beta distribution can be reduced to a function of only one parameter, thus providing a powerful tool for drawing a close approximation to the entire shape of a wing given only its centroid of area. Quite unexpectedly, the continuous spectrum of wing shapes can then be described in detail by a single parameter of shape.