496 WKITINGS OF JOSEPH HENRY. [1877 



established theory, as well as from direct observation, that 

 the sound is refracted by the wind, that it tends to be thrown 

 upward when moving against the wind, and downward with 

 the wind. This result is attributed very properly to the dif- 

 ferent velocities of the strata, that next the surface being 

 most retarded, those above being less retarded. 



The upper part of the front of the wave is thus thrown 

 backward, and the direction of the wave turned upward. 

 In the case of the experiment south of the station, the wind 

 passing over a long line of rough sea was moving less rapidly 

 in its lower stratum than in the higher, and consequently 

 the sound-wave was thrown backward above, and as it 

 issued from the instrument, tended to rise above the head 

 of the observer, and at a certain distance from the origin of 

 the sound — depending upon the difference of velocity above 

 and below, was lost entirely to the observer, and a sound- 

 shadow was thus produced by refraction, which is either sur- 

 mounted by an undulating course of the sound beam, or is 

 closed in again by the lateral spread of the sound at a given 

 distance. 



In the experiment on the other side of the signal, (the 

 vessel proceeding to the north,) the wind coming to the ob- 

 server on the vessel — had to pass over a rougher surface than 

 that of water, and consequently the difference of velgcities 

 above and below, would cause the refraction to be greater ; 

 hence the sound from the vessel was almost entirely lost to 

 the observer at the station, while the sound from the station 

 was heard uninterruptedly on the vessel, since it was mov- 

 ing with the wind. 



On examining the records of experiments of previous 

 years, I find a number of cases recorded where sounds were 

 heard at a greater distance, while inaudible at a less dis- 

 tance, especially one in connection with the fog-signal at 

 Gull Island, in 1874. In this case the sound, in passing 

 from the signal, was heard distinctly at the distance of about 

 two miles against the wind, then lost for a space of about 

 four and a lialf miles, and heard again distinctly for a dis- 

 tance of perhaps one mile. At the same station, during the 



