230 Deductions fi-om Mr. Glaisher's '' Meteorological Corrections " 



It also appears that tlie mean of IMr. Glaislier's 24-hom- cor- 

 rections, instead of being zero, is a variable quantity. For if 

 hourly observations be made and the mean taken, this would give 

 the mean daily temperature ; but if the former Avere corrected 

 sinffhj, there woidd result a mean daily temperature differing 

 from the other by the last line in my results. The diurnal for- 

 mulae may comprehend Msin/i^-l-mcosn/, wherein n=5, 7, &c., 

 or any number not an aliquot part of 24; (hoiu's). 



The mean temperature interval (Glaisher, p. 9) is for — 



h m h HI h m 



Jan. 10 April 10 20 July 12 5 

 Feb. 9 10 May 11 5 Aug. 11 



Mar. 10 10 June 12 Sept. 10 25 



Are these reducible to a formula dependent on the sun's semi- 

 diixrnal arc ? 



For the barometer, the mean epochs of the midday and morn- 

 ing means are — 



h m h tn 



Apr. + Aug.l 25 p.m. 

 May + July 1 20 

 June 1 40 

 Mean of all 1 12 



7 18a.m. 



6 33 

 4 20 



7 18 



December 45p.m. 

 Jan. +N0V.O 10 

 Feb. +Oct. 1 25 

 Mar. +Sept.l 25 



For the midday mean the above sums are pretty nearly con- 

 stant, excluding Jan. + Nov. The morning means indicate a 

 regular progression, December excepted. 



Table VII. is nearly 1'9 cos — (time from slimmer solstice, 



year =360°). 



A strict mathematical investigation would render it desirable 

 to give each monthly mean a weight corresponding to the actual 

 number of observations ; there being 26 or 27 observing days in 

 January, 24 or 25 in February, &c. ; and to take the averages of 

 four years, when the sun returns to about the same point on the 

 same day. The difference of dechnation of the odd 6 hours in 

 a common year at the equinoxes (0° 6') represents 3 or 4 days 

 at the solstices ; hence comparative weekly avei-ages would be 

 most serviceable, the average solar position corresponding to the 

 declination on the Wednesday nights. 



Mr. Glaisher's ' Sequel' (Phil. Trans. part 2, 1850) gives the 

 results for 79 years, fortunately including 4x18 and 4 x 19 

 years ; i. e. a bissextile, apogeic, and nodal cycles of the moon 

 for detecting her thermometrical influence if anywise appreciable. 



I conclude with repeating the suggestion, that an observer, 

 pro\-ided with moderately good instruments, and observing them 

 at the convenient houi's of 5, 11, 5, 11, generally equidistant 



