THE LANGUAGE OF ECHOES 



very light touches when the bat did not quite manage 

 to time its wingbeats so that the wing tips cleared the 

 obstacle. 



When the same animals not only flew through the mid- 

 dle row but also continued on through the end row, their 

 success was much less— the percentage of misses fell from 

 91 per cent at the middle row to 58 per cent for the end 

 row which was 45 centimeters from the end wall. This re- 

 sult was probably due to the very much larger echo from 

 the end wall. The situation can be understood in terms of 

 Fig. 14, a schematic graph of the sound energy reaching 

 the bat's ears during the fraction of a second when each 

 chirp is emitted and its several echoes return. The upper 

 graph (A) depicts the situation when the middle row of 

 wires is being detected and avoided; the middle graph 

 (B) applies to the same size of wire located 45 centi- 

 meters from the end wall, while the third graph (C) de- 

 scribes a further experiment in which the wires near the 

 end wall were 1.07 millimeters in diameter. In C, the 

 bat's success was about the same (88 per cent misses) 

 as it had been with the 0.46 millimeter wires at the mid- 

 dle of the flight room ( A) . Under natural conditions the 

 important echoes would be those from an insect rather 

 than a wire, and the competing echoes would arrive from 

 many different objects, such as the ground, tree trunks, 

 or branches of trees. These would produce more com- 

 pHcated echoes than those from the end wall of the flight 

 room and would be present over a longer period of time, 

 but they would never include as strong a single echo as 

 that from the large end wall. An approximation to this 

 case is represented in the fourth graph (D) of Fig. 14, 

 where it has been assumed that some of the extraneous 

 echoes have come from objects closer than the insect it- 

 self. This must happen when bats hunt, as they often do, 



101 



