156 Mr J. D. FORBES on the Horary Oscillations 



self with every precaution, and with every part of which I am 

 thoroughly acquainted. It is of the mountain construction, has 

 an adjustable level, and attached thermometer, and requires no 

 correction for capillarity. Though the other barometers were not 

 so unexceptionable, they were both furnished with attached ther- 

 mometers, and one was on the mountain construction ; indeed 

 the nature of the deductions which I have to make being entirely 

 confined to differences, without regard to absolute height, the 

 nicety of the instrument is of little importance, considering the 

 great number of observations required to be combined, in order 

 to obtain any trust-worthy result. Any constant cause, render- 

 ing the height of the mercurial column at one hour of the day 

 different from that at another, independent of the atmospheric 

 tide, must be sedulously guarded against ; but there is nothing 

 worth mentioning which can produce this effect, except change 

 of temperature. The attached thermometer must therefore be 

 considered a sine qua non. The corrections on this account are 

 applied to the monthly means ; and the table I have employed 

 is that of Professor SCHUMACHER. : it may safely be affirmed, that, 

 with rare exceptions, the tables commonly employed in this coun- 

 try are more or less erroneous, and some very strikingly so. The 

 whole of the observations are thus reduced to 32 FAHR. I may 

 also add, what is by no means unimportant, that with hardly any 

 exceptions (amounting to not above a few dozens in ah 1 ), the whole 

 of these 4410 observations were made personally by myself. 



5. In the numerical reduction of the observations, I expe- 

 rienced a very harassing difficulty. The omissions at particular 

 hours, which could not fail to occur occasionally, in the attempt 

 of an individual to register the barometer five times a-day for so 

 long a period, where they happened during extreme states of 

 pressure, would, it is easy to see, materially affect the monthly 

 means ; and although the reduction was continued in the hope 

 that the combination of a sufficient number of months would de- 

 stroy the irregularity, it became obvious, that any attempt to 



