Wind-sensitive Hairs Research Articles

Giant and dwarf axons in a miniature insect, Encarsia formosa, (Hymenoptera, Calcididae)

Miniaturization effects in the central nervous system (CNS) of a very small calchicid wasp, Encarsia formosa (0.6 mm long), are obvious for the overall morphology and at the level of axon sizes. Parasagittal sections show that most ganglia are fused and leave connectives only in the neck and the petiole. The thoracic complex is partly squeezed between muscles, enwraps cuticular apodemes and protrudes laterally into the coxae of legs. Somata of neurons are similar in size and form a multiple layer around large neuropile regions of the CNS. In TEM sections of connectives the range of axon diameters lies between 0.045 and 3.8 μm. Extremely small axon diameters below 0.1 μm are supposed to present spatial restrictions for ion channels and internal organelles. In theory, that can cause frequent spontaneous releases of action potentials (AP) which impede regular information transfer by normal APs. Therefore, axon sizes were studied in connectives between ganglia where longer distance information transfer requires action potentials even in the smallest axons. The diameters of many interganglionic axons below 0.08 μm contradict the theory. The luxury of large axon diameters exceeding 2–3 μm is reserved for several “giant” interneurons in the thoracic and in the abdominal ganglion complex. They should belong to rapid sensory alerting systems. The largest, a bilateral pair in the abdominal CNS, could integrate afferents from long wind sensitive hairs on the abdomen.

Egomotion estimation with optic flow and air velocity sensors

We develop a method that allows a flyer to estimate its own motion (egomotion), the wind velocity, ground slope, and flight height using only inputs from onboard optic flow and air velocity sensors. Our artificial algorithm demonstrates how it could be possible for flying insects to determine their absolute egomotion using their available sensors, namely their eyes and wind sensitive hairs and antennae. Although many behaviors can be performed by only knowing the direction of travel, behavioral experiments indicate that odor tracking insects are able to estimate the wind direction and control their absolute egomotion (i.e., groundspeed). The egomotion estimation method that we have developed, which we call the opto-aeronautic algorithm, is tested in a variety of wind and ground slope conditions using a video recorded flight of a moth tracking a pheromone plume. Over all test cases that we examined, the algorithm achieved a mean absolute error in height of 7% or less. Furthermore, our algorithm is suitable for the navigation of aerial vehicles in environments where signals from the Global Positioning System are unavailable.

Evolution of a new sense for wind in flying phasmids? Afferents and interneurons

The evolution of winged stick insects (phasmids) from secondarily wingless ancestors was proposed in recent studies. We explored the cuticle of flying phasmids for wind sensors that could be involved in their flight control, comparable to those known for locusts. Surprisingly, wind-sensitive hairs (wsH) occur on the palps of mouthparts and on the antennae of the winged phasmid Sipyloidea sipylus which can fly in tethered position only when air currents blow over the mouthparts. The present study describes the morphology and major functional properties of these “new” wsH with soft and bulging hair bases which are different from the beaker-like hair bases of the wsH on the cerci of phasmids and the wsH described in other insects. The most sensitive wsH of antennae and palps respond with phasic-tonic afferents to air currents exceeding 0.2 ms−1. The fields of wsH on one side of the animal respond mainly to ventral, lateral, and frontal wind on the ipsilateral side of the head. Afferent inputs from the wsH converge but also diverge to a group of specific interneurons at their branches in the suboesophageal ganglion and can send their integrated input from wsH fields of the palps and antennae to the thoracic central nervous system. Response types of individual wsH-interneurons are either phasic or phasic-tonic to air puffs or constant air currents and also, the receptive fields of individual interneurons differ. We conclude that the “new” wsH system and its interneurons mainly serve to maintain flight activity in airborne phasmids and also, the “new” wsH must have emerged together with the integrating interneurons during the evolution from wingless to the recent winged forms of phasmids.

Sensory inputs and flight orientation in locusts

1. 1. The neuronal organization of the detection and correction systems that enable locusts to flystraight are reviewed. 2. 2. Unintended deviations from course are detected by a number of exteroceptive systems including wind-sensitive hairs, compound eyes and ocelli. 3. 3. These exteroceptive signals and complementary proprioceptive signals are combined in a spatially compatible manner to generate feature detectors (the descending deviation detector interneurones) with emergent properties that relate directly to flight orientation and guidance tasks. 4. 4. The main influence of the descending deviation detector interneurones on the motor output is via intercalated thoracic interneurones, which gate the sensory signals at wingbeat frequency, thus ensuring that the corrective signals are applied at the appropriate points of the wingbeat cycle.

Synaptic Connections of Different Strength Between Wind-sensitive Hairs and an Identified Projection Interneuron in the Locust.

An identified intersegmental interneuron in Locusta and Schistocerca, with its cell body in the fourth abdominal ganglion and an axon which projects to the brain is excited by mechanosensory inputs from receptors on the head and neck. The organization of its receptive field, the types of sensory receptors which contribute to it and the patterns and strengths of the afferent connections were investigated by intracellular recording from the axon of the interneuron close to a spike-initiating site in the prothoracic ganglion. The receptive field of the interneuron consists of a small patch of hairs on the head ipsilateral to the axon, and from hairs on two regions of the prosternum (a cuticular structure on the ventral surface of the prothoracic segment), first an ipsilateral, lateral region and second a medial but contralateral region. Hairs on the pronotum (dorsal neck) also contribute but were not investigated here. Each spike in the afferent from a hair with a filiform appearance and with a pigmented base on the prosternum consistently evokes an EPSP in the interneuron. These have a short and constant latency, indicating that the connection is probably direct. The head hairs also appear to make direct connections with the interneuron in the prothoracic ganglion, so that the spike-initiating site here can integrate signals evoked by wind on the head and on the prosternum. Stiff tactile hairs on the prosternum do not connect with the interneuron. The EPSPs evoked by the long filiform hairs are consistently larger than those produced by the short filiform hairs and a single spike in some of the afferents from the long filiform hairs can evoke a spike in the interneuron. The effectiveness of an afferent is therefore correlated with the length of the filiform hair it innervates. The hairs with the most powerful effects are always the longest and occur in the same position on every locust. The shape of the receptive field and the different strengths of connections are apparent even in early larval instars. The axonal branches of the interneuron are restricted to the same side of the ganglion as the axon itself. Afferents from filiform hairs on the medial region of the prosternum project contralaterally, and those from the lateral region project ipsilaterally. Afferents from some of the head hairs project ipsilaterally directly to the prothoracic ganglion. The terminals of all these afferents overlap with the branches of the interneuron. By contrast, the afferents of tactile hairs which do not connect, project to different regions of neuropile. The connections ensure that the high sensitivity of the filiform hairs is maintained at the first stage in the central processing and suggest a role for this interneuron in supplying information about small changes in air currents that may be of use in controlling steering manoeuvres during flight.

Regional specialisation for synaptic input and output on a locust intersegmental interneurone with multiple spike‐initiating zones

A4I1 is a long intersegmental interneurone whose soma lies in the fourth abdominal ganglion and whose axon ascends through, and branches in, the thoracic and suboesophageal ganglia and the brain. It receives sensory input in the prothoracic ganglion from wind-sensitive hairs on the head and prosternum and in the fourth abdominal ganglion from cercal afferents (Pflüger: J. Comp. Neurol. 222:343-357, '84). Spikes can be initiated from both of these ganglia. In the present study the neurone was stained by intracellular injection of cobalt salts for light microscopy and of horseradish peroxidase for electron microscopy. The morphology of its neurites in the thoracic and fourth abdominal ganglia and the distribution of synapses upon them could then be observed and correlated. In the fourth abdominal ganglion the neurone receives input synapses but makes no outputs. In each thoracic ganglion the neurone has fine, varicose ventral neurites and thicker dorsal neurites. The ventral neurites are most abundant in the prothoracic ganglion where they receive only input synapses. In the meso- and metathoracic ganglia the ventral neurites also receive inputs, but in addition they make output synapses. In all three thoracic ganglia output synapses are abundant on the dorsal branches, but in the meso- and metathoracic ganglia a small number of inputs are also received here. The possibility that the function of the interneurone may not be the same in all of the thoracic ganglia is discussed in the light of its known physiological properties.

Control of hindlimb posture by wind-sensitive hairs and antennae during locust flight.

Steering movements of tethered, flying locusts, Schistocerca gregaria, subjected to simulated yaw were examined under open-loop conditions. Lateral movements of hindlimbs or curling of the abdomen were monitored with a capacitive movement transducer and were interpreted as indicating the tendency of the animal to turn. Three responses to simulated yaw were noted: Yaw-correcting upwind turning tendencies (Figs. 1, 2, 3). Downwind turning tendencies (Figs. 2, 3, 4, 5), and transient adjustments of hindlimb position consistent with an upwind turning tendency occurred in animals that made either no sustained postural adjustments of hindlimbs, or that exhibited sustained downwind turning tendencies (Figs. 4, 5). Ablations of certain mechanoreceptors tested their roles in wind detection and wind angle determination. The expression of upwind turning tendencies, whether sustained or transient, depends on inputs from cephalic mechanosensory hairplates (Figs. 2, 3, 4, 5). With hairplates occluded, all locusts exhibited downwind turning tendencies. All downwind turning tendencies depend on inputs from the antennae (Figs. 2, 3). Antennae and hairplates operate in an apparent antagonism in the steering responses they produce, which may provide the control flexibility required for complex flight maneuvering.

The development of projections and connections from transplanted locust sensory neurons

Neurons innervating wind-sensitive hairs on the locust head form characteristic projections and connections within the CNS. These depend on intrinsic properties of the epidermis from which the hair and its neuron are formed (Anderson & Bacon, 1979; Bacon & Anderson, 1984). To investigate further these intrinsic properties and also extrinsic factors involved in guiding axon growth and determining synaptic connectivity, pieces of epidermis from the head were transplanted to the posterior head, prothorax, or mesothorax. Thus wind-sensitive neurons developing from the grafts were caused to grow into foreign parts of the CNS. The neuronal projections from the graft hairs were examined by filling the axons with cobalt, and their connectivity with an identified interneuron, the Tritocerebral Commissure Giant, was examined by recording electrophysiologically the activity of the interneuron during stimulation of the graft hairs. The results show that the neuronal projections are confined to one tract, the median ventral tract, and to one arborization area, the ventral association centre, in all ganglia; in all ganglia, neurons from different epidermal regions preserve their location-specific properties of forming ipsilateral or additional contralateral projections; the extent of their projection in the CNS is not interpretable in terms of intrinsic instructions only; in foreign ganglia, they fail to form connections with their normal target interneuron.

Developmental Determination of Central Connections of Sensory Neurones From Wind-Sensitive Hairs in the Locust, Schistocerca Gregaria

ABSTRACT The nervous system of hemimetabolous insects is a favourable preparation in which to study how sensory neurones form particular projections and connections within the CNS. One reason is that the development of the sensory neurones is physically separate from the development of the CNS; sensory neurones are derived from epidermal cells under the cuticle, in contrast to interneurones and motor neurones which are derived from central neuroblasts. In addition, whereas the CNS has an almost full complement of neurones at the end of embryogenesis, neuroblast division having ceased in all but a few brain centres, sensory neurones continue to be generated from their epidermal precursors throughout the postembryonic life of the animal. Therefore transplanting pieces of epidermis during postembryonic development places sensory neurones in altered environments without surgery to the CNS. Furthermore, since sensory neurone development continues within the transplanted epidermis, the influence of these perturbations can be tested on developing as well as regenerating sensory neurones.

The large fourth abdominal intersegmental interneuron: a new type of wind-sensitive ventral cord interneuron in locusts.

A large interneuron in Locusta migratoria is described that extends from the fourth abdominal ganglion to the brain. The morphology was revealed by injection of cobaltous ions or Lucifer yellow into the cell. As its cell body lies within the fourth abdominal ganglion it is named A4I1, the first-identified intersegmental interneuron of the fourth abdominal ganglion. This neuron receives input from highly flexible, wind-sensitive hairs on the prosternum, the pronotum, and the head (field 1). Sensory connections with A4I1 are made within the prothoracic ganglion. Stimulation of the receptive field initiates spikes in A4I1 which travel anteriorly and posteriorly from the prothoracic ganglion. Intracellular recording from the axon and the soma shows that the cell membrane becomes inactive within the fourth abdominal ganglion. Spikes could be generated within the fourth abdominal ganglion by current injection into the soma. Occasionally excitatory postsynaptic potentials were observed in a soma recording, but up to now there is no evidence for a second spike initiation site. By intracellular current injection into the soma of the left and right A4I1 cell it is shown that the two cells are not electrically coupled.

Short Communications Initiation of Flight By an Identified Wind Sensitive Neurone (Tcg) in The Locust

ABSTRACT When the tarsi are not in contact with the ground, brief wind stimuli applied to the wind sensitive hairs on the head elicit flight in the locust (Weis-Fogh, 1949; Camhi, 1969). The descending interneurones mediating this reaction have not been identified but some likely candidates have been reported (Bacon & Tyrer, 1978; Simmons, 1980, 1981). The most extensively studied is the tritocerebral commissure giant (TCG) (Bacon & Tyrer, 1978; Bacon & Möhl, 1979). This large wind sensitive interneurone descends from the brain, crosses to the contralateral side of the nerve cord via the tritocerebral commissure, and makes excitatory connections with some flight motor neurones in the thoracic ganglia (Bacon & Tyrer, 1979). The location of the TCG’s axon in the tritocerebral commissure makes this neurone easily accessible for recording and stimulation in almost intact, freely moving animals. Recordings during flight, for example, have shown that the TCG discharges one or two spikes during each cycle (Bacon & Möhl, 1979). The spike activity results from rhythmic air turbulences caused by flapping of the wings and head movements during flight. This rhythmic activity may play a role in regulating motor output during flight manoeuvres, such as yaw (B. Möhl & J. Bacon, in preparation). It seems unlikely however that this is the only function for the TCG since it is capable of discharging in very high frequency bursts quite unlike those occurring in flying animals. Here we present evidence that the TCG also functions to facilitate the initiation of flight following a jump.

The phasic influence of self-generated air current modulations on the locust flight motor

1. Using high speed film analysis (500 frames/s) to investigate the head nodding movement during tethered flight inLocusta shows that the position of the wind-sensitive head hairs with respect to the flight direction is altered by 5.5° in the rhythm of the wing beat (Fig. 2). 2. Wind measurements in the region of the hair fields demonstrate that the wind reaching the hairs during flight is modulated by the animal's own wing beat. The modulation has a peak-to-peak value of 0.6–1.0 m/s (Fig. 3). 3. An airstream with its speed modulated by these values was used to stimulate the wind-sensitive hairs to analyse the steady-state response during tethered flight in animals with the antennae removed (Fig. 1). In these entrainment experiments absolute coordination (a relation of locked phase) between the wind modulation and the flight oscillator is found in a range of about 3 Hz around the intrinsic flight frequency. At frequencies both above and below this range, relative coordination (a relation of preferred phase) is obtained (Figs. 4–7). 4. The dynamic response to step changes in the modulation frequency was tested. There is an immediate reaction, but it takes several wing beats to reach the new steady-state (Fig. 8). 5. When flight was elicited while a modulated wind stream was already blowing, the first wing beat occurred in a preferred phase with respect to the stimulus modulation (Figs. 10 and 11). 6. To understand the generation of the flight pattern, the whole flight oscillator must be considered as a cooperative system of central neuronal, sensory (proprioceptive) and mechanical components (Fig. 12).

Parallel encoding of direction of wind, head, abdomen, and visual pattern movement by single interneurons in the dragonfly

1. The self-movement detectors whose detailed properties are introduced here are among the largest interneurons in the ventral nerve cord of the dragonflyAnax junius. They combine information from a wide variety of sensory structures including the compound eyes, mechanosensory frontal hairs, antennae, neck proprioceptors, and wind sensitive hairs in the neck region (Fig. 6). They fall into 8 classes, each with a specific pattern of multimodal inputs (Fig. 1). 2. Their visual receptive fields are large, anteriorly weighted, and selective for animal rotation in a given direction (Figs. 1, 2). 3. They show directionally selective responses to air puffs delivered to the front of the head and to shifts in direction of a stream of air onto the head (Fig. 8). 4. They show directionally selective responses to air puffs from the sides, above, and below the animal into the neck region (Figs. 3, 4, 5). 5. They respond directionally to rotations of the head in still air in the dark (Figs. 3, 4, 11). The receptors mediating these proprioceptive responses are probably different from those mediating the responses to air puffs to the neck. 6. Four of the eight classes respond when the animal actively flexes its abdomen (Fig. 12). 7. Within each modality, they are inhibited by stimulation in the non-preferred direction. Pairwise presentations of stimuli to different inputs showed that each sensory channel is capable of inhibiting self-movement-detector responses to stimulation through any other channel (Figs. 13, 14). 8. For each class, the receptive fields of the various inputs are all aligned to detect rotation of the animal or the head within the same plane in space (Figs. 3, 4, 5). Two classes are aligned in the horizontal plane, and the other six in oblique planes (Fig. 1). 9. The precise alignment of directional selectivity of several input modalities within the same plane, combined with the finding that individual self-movement detectors drive adjustments in wing position (Olberg 1978) indicates that these interneurons are critical elements of a feedback control system for flight control.

Synaptic transmission between second- and third-order neurones of a locust ocellus

1. This paper describes the operation of synapses which large, second-order neurones of each ocellus of a locust (L-neurones) make with a pair of large, identified descending neurones on either side of the brain (Fig. 1). Intracellular recordings were made from the axons of the L-neurones, and usually from the cell bodies of the third-order neurones. 2. The third-order neurones respond, in addition to changes in illumination, to currents of air which deflect wind sensitive hairs on the head (Fig. 2B). 3. Both L-neurones and the third-order neurones hyperpolarise when their ocellus is illuminated (Figs. 2, 3). The hyperpolarising responses of the third-order neurones saturate at a light intensity which is well within the range of intensities to which the L-neurones respond (Fig. 3). L-neurones and third-order ocellar neurones produce sharply rising regenerative responses when a bright light is switched off, and the third-order neurones spike at less intense changes in illumination than L-neurones do (Fig. 3A). 4. L-neurones make excitatory chemical synapses with the third-order ocellar neurones (Figs. 4, 5). In steady daylight illumination, L-neurones continually release transmitter onto the third-order neurones (Fig. 4). The hyperpolarising responses of the third-order neurones to increases in illumination are due to decreases in the rate of release of transmitter from the L-neurones.

A Locust Wind and Ocellar Brain Neurone

ABSTRACT One of the large descending brain neurones in each half of the central nervous system of a locust is probably important in regulating the attitude of the locust’s flight in the pitch plane. This function is suggested by a consideration of the stimuli which excite the interneurones, and of the muscles whose motoneurones they excite. Each of these interneurones has its cell body in the protocerebrum and its axon descends the ipsilateral connective and has branches in all of the thoracic ganglia. The interneurones are excited by currents of air directed at the front of the head. An increase in the intensity of light falling on the median ocellus causes a reduction in the frequency of spikes produced by stimulation of the wind-sensitive hairs, and a reduction in the intensity of this light causes an increase in the frequency of these spikes. Each interneurone makes excitatory connexions with motoneurones of the subalar and dorsal longitudinal flight muscles ipsilateral to it.

Sensory projections from the wind-sensitive head hairs of the locust Schistocerca gregaria. Distribution in the central nervous system.

The neurones from the wind-sensitive hairs on the locust head have been filled with cobalt chloride and intensified with silver. All the neurones project through the brain to the suboesophageal ganglion, some continue to the prothoracic ganglion and a few as far as the mesothoracic ganglion. Three different types of projection are described and a regrouping is proposed of Weis-Fogh's five hair fields into three areas. The distribution of the neurones from these areas is described in relation to other structures in the ganglion and is discussed in relation to the function of the hair fields in stability control and grooming.

Developmental determination of neuronal projection patterns from wind-sensitive hairs in the locust, Schistocerca gregaria

Neurons innervating wind-sensitive hairs on the head of Schistocerca gregaria send axons to the suboesophageal ganglion where they form one of two distinct projection patterns. Transplantation experiments show that the type of pattern formed depends upon the origin of the epidermis from which the hair and its associated neuron are derived.

Suboesophageal neurons involved in head movements and feeding in locusts.

The projections of nerves 6 and 7 of the locust suboesophageal ganglion (SOG) were stained by axonal filling with cobalt chloride. Nerve 6 contains two motoneurons which innervate neck muscles 50 and 51. Sensory neurons innervating hairs on the dorso-occipital region of the head also enter the ganglion through nerve 6 and terminate in a small bilateral plexus. The projections of the head hairs in nerve 6 do not overlap the arborizations of the motoneurons or the neurons of nerve 7, but lie in the same area as descending sensory neurons from wind-sensitive hairs of the front of the head. One branch of nerve 7 (7B) contains two fibres which innervate the salivary gland. These 'salivary' neurons (labelled SN1 and SN2) have their cell bodies in the ganglion. The second branch, 7A, contains sensory neurons from the submentum of the labium, which form four sensory plexuses, two dorsal and two ventral. The sensory plexuses from the submentum have specific regions of overlap with the salivary neurons and with the neck muscle motoneurons. We interpret these as indicating a flow of information from labial receptors signalling head and mouthpart movement to neurons involved in salivation and head movement. We further postulate that the anatomical separation of the various sensory plexuses is indicative of functional localization within the ganglion.

The influence of the air-current sense organs on the flight behaviour ofLocusta migratoria

1. Several flight parameters, the wing-beat frequency, the wing-stroke angles, the lift and the flight speed, were measured simultaneously by means of a pendulum or a transducer apparatus (Figs. 1, 2) in tethered flyingLocusta migratoria (♀♀). 2. In normal locusts all these parameters decrease with increasing duration of flight (Figs. 3–6), i.e., they are positively correlated (Figs. 8, 9). The wing-beat frequency seems to be the primary factor in governing flight speed and lift. 3. It is demonstrated by elimination experiments that all measured flight parameters are controlled in part by the air-current sense organs of the head, i.e., the hair patches and the antennae. After immobilization of the wind sensitive hairs the wing-beat frequency and the flight speed are reduced with respect to normal animals (Figs. 3c, 4c). If the antennae are eliminated, however, the wing-beat frequency, the wing-stroke angle and the flight speed are increased (Figs. 3b, 4b, 5a, c, 6a). These findings suggest that in free flying locusts the air-current sense organs affect the flight speed in opposite ways: the hair patches stimulate it and the antennae reduce it, they being the sensory units of a negative feedback mechanism. 4. Since the flight speed is increased and the lift decreased in flagella-less locusts in relation to normal ones, it is clear that the antennae govern the direction of the “resulting force” of these tethered flying locusts (Fig. 10b). 5. The maintenance of flight is enhanced by both the hair patches and the antennae.

Biology and physics of locust flight IV. Notes on sensory mechanisms in locust flight

The wing stroke of locusts is remarkably constant and independent of external conditions. Is this rigid rhythmicity due to a rhythmicity of the central nervous system or is it determined by peripheral factors? The flight behaviour of the desert locust ( Schistocera gregaria ) was studied under various experimental conditions in order to find which external factors can initiate, maintain or alter the wing movements, excluding reactions which depend upon higher nervous centres. The ‘tarsal reflex’ and the response seen when the aerodynamic sense organ on the head is stimulated (Weis-Fogh 1949, 1950) were reinvestigated in order to relate them to two hitherto unknown reactions: the maintenance of flapping when the wings are exposed to wind and the regulation of the lift when the body angle ( = angle of pitch) is changed during forward flight. Both depend on receptors whose nature is still unknown. Inhibition . As in most other insects, the flight of a locust cannot be started when the legs, or only part of one leg, have contact with a rigid body; flight stops when such contact is regained. Amputation of the legs abolishes these reactions, showing that some leg proprioceptors inhibit flight. Initiation . A suspended locust can be induced to fly in three ways. (1) By application of a sufficiently strong stimulus which normally provokes escape reactions; the flight lasts only a few seconds. Adaptation is generally quick. (2) By sudden removal of the support for the legs (‘ tarsal reflex’ although not confined to the tarsi). The flight lasts 5 s on average, corresponding to one hundred wing strokes. There is practically no adaptation. (3) By blowing upon the wind-sensitive hairs on the head. The wind must exceed 2 m/s, but its direction is of little importance. Since the static bending has no effect, the adequate stimulus seems to be minute vibrations of the hairs. The flight lasts as long as the wind blows and the hairs are therefore also involved in the maintenance of flight. When the locust has stopped, the legs begin to flutter, and eventually remain still, but normally flight is not resumed unless one of the above stimuli is applied. Maintenance . Two receptor systems are involved. (1) The wind-sensitive hairs on the head. In a wind they emit impulses irrespectively of whether the locust has any chance of flapping its wings or not. ‘Wind on the head’ is therefore an extrinsic flight stimulus. The flight posture is never complete. (2) A hitherto unknown receptor system in the pterothorax which was studied in insects whose supra-oesophageal ganglia were cauterized (‘decerebrate’). It maintains the movements when the wings oscillate in a wind but cannot initiate them; the adequate stimulus is the rhythmically changing wind pressure on the wings. ‘Wind on the wings’ is therefore an intrinsic flight stimulus. When the average lift exceeds half the body weight, flight continues in complete flight posture but stops when the lift approaches zero. The experiments indicate that the stimulation ceases when the lift becomes negative during the upstroke . The receptors are unknown; it is suggested that they are situated at the wing hinge. The locust does not adapt to either of these stimuli and invariably stops a few seconds after they have ceased. Control of lift . The locust tends to keep the lift constant during a given performance. This observation, together with the constancy of most stroke parameters, made it possible to investigate the mechanism involved. The method was to make the insect fly steadily against a horizontal wind and then alter the inclination to the wind (= the body angle) at regular intervals. The data permitted an estimate of the mean change in wing twisting Δθ. Δθ increased (wings pronated) by 15 ± 3° when the body angle was increased from 0 to 15°. This is the main factor in the control of lift. The discussion shows that this presupposes a system of lift-sensitive receptors (probably campaniform sensilla at the wing hinge). If present in other insects, the homoeostatic character of the wing stroke of Drosophila (Chadwick 1953) may therefore be caused by a nervous mechanism and need not be a consequence of the energetics of flight. Central rhythm . It is concluded that the central nervous system ( does not initiate flight rhythm de novo ; ( b ) does neither determine the stroke frequency nor the strength of the contractions of the controller-depressor muscles; ( c ) may control the phasing of the contractions, although a simpler hypothesis is advanced; ( d ) may control the indirect flight muscles but only as far as to produce stimuli of constant (maximum?) strength.