RECENT EXPERIMENTS ON FOG-SIGNALS 291 



cance. There was no visible reflecting surface from which 

 they could come. On some days, with hardly a cloud in 

 the air and hardly a ripple on the sea, they reached a 

 magical intensity. As far as the sense of hearing could 

 judge, they came from the body of the air in front of the 

 great trumpet which produced them. The trumpet blasts 

 were five seconds in duration, but long before the blast had 

 ceased the echoes struck in, adding their strength to the 

 primitive note of the trumpet. After the blast had ended 

 the echoes continued, retreating further and further from 

 the point of observation, and finally dying away at great 

 distances. The echoes were perfectly continuous as long 

 as the sea was clear of ships, "tapering" by imperceptible 

 gradations into absolute silence. But when a ship hap- 

 pened to throw itself athwart the course of the sound, the 

 echo from the broad side of the vessel was returned as 

 a shock which rudely interrupted the continuity of the 

 dying atmospheric music. 



These echoes have been ascribed to reflection from the 

 crests of the sea-waves. But this hypothesis is negatived 

 by the fact that the echoes were produced in great inten- 

 sity and duration when no waves existed when the sea, 

 in fact, was of glassy smoothness. It has been also shown 

 that the direction of the echoes depended not on that of 

 waves, real or assumed, but on the direction of the axis 

 of the trumpet. Causing that axis to traverse an arc of 

 210, and the trumpet to sound at various points of the 

 arc, the echoes were always, at all events in calm weather, 

 returned from that portion of the atmosphere toward which 

 the trumpet was directed. They could not, under the cir- 

 cumstances, come from the glassy sea; while both their 

 variation of direction and their perfectly continuous fall 



