550 RESEARCHES IN SOUND. 



tbat the automatic buoy is found to intermit its sound, being heard at 

 a distance, then becoming inaudible, and heard again as the steamer 

 approaches the source of sound. 



From all the facts which we have gathered on this subject, I think it 

 highly x^robable that in all cases in which sound moving against the 

 wind is thrown up above the head of the observer it tends to descend 

 by the lateral spread of the sound-wave and to reach the earth at a dis- 

 tance; the conditions however for the actual production of this effect 

 are somewhat special, and will depend ui)on the amount of the initial 

 refraction and the quantity of the sound-waves. Besides the lateral 

 spread of the sound-wave there are two other causes sufficient, in certain 

 cases, to biing a portion of the sound-waves which have been elevated 

 in the air back again to the earth : the first is when an upper current 

 of wind is moving in an opposite or approximately opposite direction 

 to that at the surface of the earth, in which case an opposite or down- 

 ward refraction would take place ; and the second is the case in which 

 the siuface-wind is terminated above by strata of still air ; in this case, 

 also, a reverse refraction, but of less amount, would take jjlace, which 

 would tend to bring the sound-wave downward. 



We can readily imagine that an isolated island, cooled by the radia- 

 tion of the heat by night, would send every morning, in all directions, 

 a current of cold air from its center. In this case, the sound from a 

 whistle placed in the center of the island would be inaudible in a space 

 entirely surrounding it, and thus give rise to a condition mentioned by 

 General Duane, in which a fog-signal aijpeared to be surrounded by a 

 belt of silence. 



The next experiment was made on the morning of the 5th, on leaving 

 the station. In this case we proceeded along the direction of the same 

 line in which the first, second, and third experiments were made the day 

 before. The wind had changed about four points to the southward. 

 As in the preceding experiments, the sound was lost again at the dis- 

 tance of about one-fourth of a mile, but was not distinctly regained, 

 though some of the observers thought they heard it at a distance of two 

 and one-half miles. 



The only x)erceptible difference in the wind on the 5th was that it was 

 a little less rapid, and four points more to the southward. 



From a subsequent report of the keepers, the whistle of the vessel 

 was heard continuously as far as the i^ufis of steam could be observed, 

 a distance six or seven miles. In this case the sound was moving with 

 the wind. These results therefore are in accordance with those pre- 

 viously obtained. 



The next experiments were made at Monhegan, an island sixteen miles 

 southwest of White Head. On this island there is a Daboll trumpet 

 actuated by a hot-air engine. 



We departed from this station in. a westerly dkection at an angle of 



