278 



Sir G. B. Airy. On the 



[Nov. 23 y 



labour of making observations at every hour of day and night, the 

 conclusions deduced from those few instances might have been 

 vitiated by accidents. But I am able to assert positively, as a result 

 from the reductions to be exhibited in the following pages, that 

 nothing has been lost from the restriction of the plan of observation. 

 It will be seen that the daily change of temperature, produced by the 

 aggregate of strictly diurnal change (depending on the solar hour) 

 and tidal change (depending on the moon's apparent position) is so 

 small, that it is impossible to attach with any certainty a sensible 

 value to either of these causes. 



I now proceed to describe the principal steps in the reduction of the 

 observations. 



In the weekly publication of these observations by the Registrar- 

 General, the weekly means of each observed element were also 

 exhibited. In preparation for a detailed publication of the whole, 

 I had the entire series of these weekly means collected, each being 

 accompanied with notes of the principal phases of the moon, the 

 occurrence of remarkable storms, &c, occurring within the week. 

 (This resume exists, and is available for any discussion which might be 

 suggested ; I propose to offer it for deposit at the Royal Observatory.) 

 But on general examination of the collected means, I did not perceive 

 that any result could be expected which would justify the labour and 

 expense of printing the whole. For instance, if there were any 

 remarkable dependence on the phase of tide, different values for the 

 " excess of mean temperature of the water above mean temperature 

 of the air," would occur in the weeks which included respectively 

 new moon, first quarter, full moon, third quarter ; and these would 

 recur with little alteration for several months. But on general 

 examination I do not see anything which would justify more technical 

 discussion directed to this point. Finally, I decided on exhibiting 

 only the means of deductions as to temperature for each calendar 

 month, and omitting all other phenomena. As the succession of 

 weeks and the succession of entire months do not generally coincide, 

 the rule was established, to adopt the first entire week in each 

 calendar month as the first of the weeks to be used, in conjunction 

 with three or four weeks following, to form the monthly mean. 

 Thus, some months contain four weeks and some contain five weeks. 

 For instance, the month of 1846, March, contains the five weeks, 

 March 1—7, 8—14, 15—21, 21—28, 28— April 4; but the next 

 month contains only the four weeks, April 5—11, 12 — 18, 19 — 25, 

 26— May 2. 



By this system, the results, as far as they appear to possess any 

 value, are brought into the compass of five convenient tables of 

 double entry, which, with their columnar and lateral means, appear 

 to give all the information that can be desired. The contents of the 

 several tables are : — 



