632 Royal Society .— 



sky. Hence it is shown that " rays " of sound, otherwise hori- 

 zontal, will be refracted upwards in the form of circles, the radii 

 of which are 110,000 feet with a clear skv, and 220,000 with a 

 cloudy sky— that is to say, the refraction on bright hot days will 

 be double what it is on dull days, and still more under exceptional 

 circumstances, and comparing day with night. 



It is then shown by calculation that the greatest refraction 

 (110,000 radius) is sufficient to render sound, from a cliff 235 feet 

 high^ inaudible on the deck of a ship at 1| mile, except such sound 

 as might reach the observer by divergence from the waves passing 

 over his head; whereas, when the refraction is least (220,000 

 radius), that is, when the sky is cloudy, the range would be ex- 

 tended to % 2\ miles, with a similar extension for the diverging waves, 

 and under exceptional circumstances the extension would be much 

 greater. It is hence inferred that the phenomenon which Prof. 

 Tyndall observed on the 3rd of July and other days (namely, that 

 when the air was still and the sun was hot he could not hear guns 

 and other sounds from the cliffs 235 feet high more than 2 miles, 

 whereas when the sky clouded the range of the sounds was exten- 

 ded to 3 miles, and, as evening approached, much further) was due, 

 not to the stoppage or reflection of the sound by clouds of invisible 

 vapour, as Prof. Tyndall has supposed, but to the sounds being lifted 

 over his head by refraction in the manner described ; and that, had 

 he been able to ascend 30 feet up the mast, he might at any time 

 have extended the range of the sounds by a quarter of a mile at 

 least. 



April 30. — Prof. Andrew Crombie Ramsay, I/L.D., Vice-President, 

 in the Chair. 



The following communication was read : — 



" On the Improvement of the Spectroscope." Bv Thomas Grubb, 

 F.K.S. 



The importance, as an instrument of research, which the spec- 

 troscope has reached within a few years, renders any improve- 

 ment therein a matter of general scientific interest. Hitherto it 

 has been under a disadvantage, which, though slight in amount 

 in those cases in which the dispersive power of the instrument is 

 moderate, becomes a rather serious annoyance to the observer when 

 a number of prisms are used in serial combination, and the cur- 

 vature of the spectral lines is proportionally increased, and only 

 to be restrained in appearance by using a narrow breadth of the 

 spectrum. 



I have lately thought of a very simple and practical remedy 

 (which may indeed have occurred to others, but which I have not 

 seen mentioned), whereby those lines are rendered palpably 

 straight in a very large field : but previous to describing it, it is 

 desirable to refer to a statement appearing in the ' Astronomical 

 Xotices ' for last month (March), viz. that the spectral lines can 





