ABSTRACT
The extension mechanism of the larval dragonfly’s labium has been under discussion since Snodgrass (1954) drew attention to the absence of muscles at the head/postmentum joint (p–h, Fig. 1A) and proposed that it was extended by blood pressure, due to the abdominal muscles which also cause jet propulsion. This proposition is supported by evidence that the anus closes (Pritchard, 1965), and the abdominal muscles contract (Caillère, 1972; Olesen, 1972) during labial extension; and by reports of an increase in abdominal pressure of 5000 Pa (Olesen, 1979) and 4000—12000Pa (Tanaka & Hisada, 1980). From high-speed film of Aeschna nigroflava, Tanaka & Hisada (1980) calculate a maximum torque of 40 αN.m (neglecting water resistance) and state that this would require a joint pressure of 6000 Pa. They also claim that the labium extends normally when a pressure of 8000 Pa is applied to a dead animal under water, thus indicating the relatively small effect of water resistance. However, these conclusions raise a number of questions. In deriving the pressure of 6000 Pa, they appear to have used the torque/pressure characteristic of the postmentum/prementum joint (p–p, Fig. 1A). Had they used the p-h joint characteristic (their Fig. 6) a required pressure of 24 000 Pa would have been obtained. And the statement that 8000 Pa, applied to a head preparation under water, produces normal extension, is not borne out by a comparison of their Figs 5 and 8. which again only relates to the p–p joint. Confusion between the p–h and p–p joint may be due to the erroneous view that the torques at the two joints must, for mechanical reasons, be equal. Their data on labial movement is subject to large standard error and no estimate is made of the error caused by the use of a simplified model, which assumes that the mass of the prementum is concentrated at the end of the postmentum.
I have used fresh data, from Anax imperator (Leach), and a less approximate model, to recalculate the p—h joint torque and estimate the effect of water resistance; and have also briefly considered the p–p joint extensor mechanism. Second-year larvae were filmed at 500 frames s–1 and 15-20 °C while attacking living Calliphora grubs (Fig. 2). From a number of trials, three successful sequences were obtained involving two specimens of different size. From plots of the angles θ and ϕ (Figs 1A and 3), angular velocities (ω) and accelerations (α) were calculated over 4-ms intervals.
If, instead, as assumed by Tanaka & Hisada (1980), the mass of PQ is concentrated at P, this expression reduces to: TO● /l25=14·5 αθN.M/m5. Neglecting water resistance, maximum values for TO /l25, and the corresponding torques To, for the three sequences analysed, were:
Tanaka & Hisada (1980, Fig. 6) obtain a torque/pressure ratio equivalent to 0-0023 I23 N.m/Pa for the p-h joint of A nigroflava. Now torque/pressure α AV/A θ where V and θ are joint volume and angle (Parry & Brown, 1959) and as the anatomy of the joints in this species and in A. imperator are similar, I have applied the same ratio to the latter, giving the pressures shown above. These pressures are well within the maximum values recorded. The much greater torque and corresponding pressure in A. nigroflava is mainly due to its much more massive prementum (m2 = 60·4/23 kgm–3 c.f. 24·7/23kgm–3 in A. imperator). Note that concentrating the mass of the prementum at the end of the postmentum may over-or underestimate the maximum torque depending on the exact configuration of the extending labium.
where s and l are the width and length of the labial segment, and Q is the density of water (Lighthill, 1975, p. 74); and assume that its centre of mass and radius of gyration coincide with that of the corresponding labial segment. We find that mw/l23= 26·1 and 121·1 kgm–3 for the postmentum and prementum respectively, and: 
Tanaka & Hisada (1980) attach importance to a click mechanism at the p–p joint of Aeschna nigroflava — viz. paired knobs on the postmentum that slide along grooves in the prementum ‘as if the engaged position is the real pivot of the p—p joint’ (which would be inconsistent with a firm joint articulation). In Anax imperator no such structure has been found and the joint may be observed to rotate smoothly about its articulation.
ACKNOWLEDGEMENTS
The high-speed film was taken by Mr Gordon Runnals, and specimens were collected with the help of Drs Janet and Norman Moore. I am also most grateful to Professor Charles Gurney and Dr K. E. Machin for their advice on the mechanical analysis, and to Dr Machin for his comments on the manuscript.
