RESEARCHES IN SOUND. 549 



abnndant testimony it is establislied tliat tbe same plienomeua are ex- 

 liibited during a dense fog, in wbicb all parts of tbe atmospbere for 

 miles in extent must be in a bomogeneous condition. Furtbermore, a 

 local cloud could not continue to exist in a given space for more tban an 

 instant wbile a wind was blo^ying witb a velocity of from ten to twelve 

 miles an bour. Again, tbis bi^-potbesis fails entirely to explain tbe fact 

 tbat tbis pbenomenon is always observed at nearly tbe same place, espe- 

 cially during a fog, wben tbe wind is in a soutberly direction. Finally, 

 it is impossible to conceive of a cloud so arranged as a screen produc- 

 ing a sound-sbadow of greater intensity on one side tban on tbe otber. 



Anotber bypotbesis is tbat of tbe refraction of sound due to tbe action 

 of tbe wind. It is an inference from well-establisbed tbeory, as well as 

 from direct observation, tbat tbe sound is refra^jted by tbe wind, tbat it 

 tends to be tbrown u^iward wben moving against tbe wind, and down- 

 ward witli tbe wind. Tbis result is attributed very properly to tbe dif- 

 ferent velocities of tbe strata, tbat next tbe surface being retarded, 

 tbose above being less retarded. 



Tbe upper part of tbe front of tbe wave is tbus tbrown backward, and 

 tbe direction of tbe wave turned upward. In tbe case of tbe experi- 

 ment soutb of tbe station, tbe wind passing over a long Hne of rougb 

 sea was moving less rapidlj^ in its lower stratum tban in tbe bigber, and 

 consequently tbe sound-wave was tbrown backward above, and, as it 

 issued from tbe instrument, t-ended to rise above tbe bead of tbe observer, 

 and at a certain distance from tbe origin of tbe sound, depending upon 

 tbe difference of velocity above and below, was lost entkely to tbe ob- 

 server, and a sound-sbadow was tbus j)roduced by refraction wbicb is 

 closed in again by tbe lateral spread of tbe sound at a given distance. 



In tbe experiment on tbe otber side of tbe signal, tbe vessel proceed- 

 ing to tbe nortb, tbe wind coming to tbe observer on tbe vessel bad to 

 pass over a rougber surface tban tbat of water, and consequently tbe 

 difference of velocities above and below, and tberefore tbe refraction 

 would be greater, and consequently tbe sound from tbe vessel was almost 

 entirel}" lost to tbe observer at tbe station, wbile tbe sound from tbe 

 station was heard uninterruptedly on tbe vessel, since it was moving 

 witb tbe wind. 



On examining tbe records of experiments of previous years, I find a 

 number of cases recorded wbere sounds were beard at a greater dis- 

 tance, wbile inaudible at a less distance, especially one in connection 

 witb tbe fog-sigual at Gull Island, in 1874. In tbis case tbe sound, in 

 j^assing from tbe signal, was beard distinctly at tbe distance of about 

 two miles against tbe wind, tben lost for a space of about four and a 

 balf, and beard again distinctly' for a tlistance of perbaps one mile. At 

 tbe same station, during tbe experiments of 1875, tbe sound of the 

 wbistles of tbe steamers was beard for a certiiin distance, tben ceased 

 to be beard for a considerable interval, and was tben beard again. Fur- 

 tbermore, tbe i)ilots of tbe steamboats from Xew York to Boston report 



