RELATION BETWEEN PREY AND PREDATOR — ROEDER 295 



this Study yielded the following time intervals between loss of tarsal 

 contact and the onset of active flight movements: Phormia regina at 

 i8° to 25° C, 45, 50, 50, 60, and 65 msec; an unidentified tabanid, 

 55, 65, and 85 msec ; Eristalis sp., 45, 60, and 65 msec. A few measure- 

 ments with Hymenoptera gave much more variable flight reflex times, 

 a warmup period being necessary under certain conditions in this 

 order. The shortest time obtained was 70 msec, in an individual honey 

 bee at 25° C. With other bees and with Polistes the time was often 

 as long as 300 to 400 msec, and flight could not be initiated at all by 

 this method in many instances. 



An attempt was made to measure the startle response time of the 

 cockroach, Periplancta amcricana. Adult males were used and the 

 experiments were carried out at room temperature. A light wooden 

 support was fixed with wax to the pronotum and inserted into a 

 peizoelectric phonograph pickup. The feet of the insect rested on a 

 freely movable paper platform so that any sudden backward thrust 

 of the legs was registered as a forward thrust by the support and 

 pickup. A second pickup was mounted near the cerci so that it 

 registered the arrival of a puff of air directed at this region. The time 

 interval elapsing between deflection of the second pickup by the air 

 blast and the deflection of the pickup attached to the pronotum was 

 measured by recording the electrical output of both on a cathode-ray 

 oscilloscope. 



The insects were restless under this restraint, and chance move- 

 ments invalidated many of the measurements. Another source of 

 error lay in the uncertainty of what constituted a critical level of 

 stimulus as the air blast grew in intensity. Twenty-three measure- 

 ments gave an average value of 54 msec, and a range of 28 to 90 msec. 

 Resolution of the startle response of the cockroach into its con- 

 stituent neural events was next attempted. Previous studies (Pum- 

 phrey and Rawdon-Smith, 1937; Roeder, 1948) have shown that 

 mechanical displacement of lightly balanced hair sensilla on the cerci 

 generate afferent impulses in the cereal nerve fibers. The latter form 

 synapses in the last abdominal ganglion with the giant fiber system 

 mentioned above. The giant fibers ascend the abdominal nerve cord 

 without further synapses although they decrease in diameter and con- 

 duction is slowed as they pass through the neuropile of each abdominal 

 ganglion. On reaching the metathoracic ganglion the giant fibers form 

 unstable synapses with the system of motor neurons innervating the 

 metathoracic legs whence they continue up the nerve cord to form 

 similar connections in the prothoracic and mesothoracic ganglia 



(pl- 5). 



