1886.] The Harmonic Analyser at the Meteorological Office. 389 



At present only the monthly increments of the readings have been 

 dealt with, so as to obtain the coefficients of the mean daily variation 

 for each month of the year. The process followed is, therefore, 

 simply to divide the monthly increment by the number of days in the 

 month, and then to multiply the quotient by a factor which is deter- 

 mined by the scale-value of the thermograms, and which will there- 

 fore be different for each observatory. 



The ratios of the factors for cylinders 1 to 6 to that of No. 7 

 were very carefully determined from a series of experimental curves, 

 of which the values were known. The numerical factor is obtained 

 for each observatory by obtaining on cylinder No. 7 the scale reading 

 corresponding to a vertical movement of the pointer of 10° on the 

 thermogram, which in the case of Kew is 075 inch. The factor for 

 cylinders 1ST os. 1 and 2 is eight times that for cylinder No. 7 ; the 

 factor for Nos. 3 and 4 is four times that quantity, and for Nos. 5 and 

 6 is eight-thirds of that quantity. The signs of the factors depend on 

 the direction in which the disks and cylinders are caused to revolve. 

 The constant quantity added to the reduced reading of cylinder No. 7 

 corresponds to the temperature which is assumed as the zero at the 

 commencement of the operation. 



As an illustration, the case of Kew for July, 1882, may be taken, 

 the final readings of the cylinders for which month are above given. 

 The increments for the month shown by these figures are as fol- 

 lows : — 



Cylinder 1. 2. 3. 4. 5. 6. 7. 



°mTnT d .. m ? re ."} + 2 ' 198 ~ 2 ' 671 -0-101 -0-198 -0-797 -0'564 +56'839 

 Divided by 31] 



(the number f f0'07l -0-086 -0/003 -0-006 -0*026 -0*018 + 1-834 

 of days) .... J 



Factor -5352 + 53-52 -26*76 -26'76 -17 '84 -17-84 + 6*69 



C °dS n !. . d . e ."} ~ 3 ' 80 ~ 4 ' 60 + °' 08 + °' 16 + °' 46 + °' 32 + 12 ' 2 * 



Add constant 48 - 17 



Mean temperature 60 *44 



After some trials with the curves for the year 1871, the year 1876 

 was taken up, inasmuch as for that year the records had been dis- 

 cussed by Mr. H. S. Eaton, M.A., F.R. Met. Soc, who had calculated 

 the hourly means of the various meteorological elements for each 

 month separately, and who kindly placed his results at the disposal 

 of the Council. 



The working of the machine was thus subjected to an exact test by 

 comparing the results obtained by it with the coefficients in the har- 

 monic series which were calculated from Mr. Eaton's means ; and 

 their trustworthy character, and the adequacy of these calculations to 



