1886.] The Harmonic Analyser at the Meteorological Office. 391 



to failure of photography, &c, were interpolated when the curves 

 were passed through the machine, but not when the means were 

 worked out from the hourly measurements. Some differences rather 

 larger than usual, and confined chiefly to the earliest years dealt 

 with, have been ascertained to have arisen from the circumstance that 

 when the curves were first measured, to obtain hourly values, the 

 method of making the measurements was not the same as that found 

 by subsequent experience to be the preferable ; and also that in some 

 cases the scale- values first used were less accurately determined than 

 has since been found possible. 



In both these respects the two methods were on a par in the later 

 years dealt with, and therefore the fairest comparison is to be had 

 with their means. 



For 1880, the average difference of the monthly mean for all the 

 seven observatories is 0-09° ; for 1881 it is 0-05° ; and for 1882 O06° ; 

 and in these three years a difference of 0*3° between the analyser 

 and calculated means occurred but once, and of 02° but five 

 times. 



What has been said is sufficient to show that the instrument is 

 completely applicable to the analysis of thermograms. 



It has also been employed on the discussion of barograms, and the 

 curves for the years 1871, 1872, and 1876 have been passed through 

 the machine. 



The year 1876 was selected owing to the existing facilities for 

 comparing the resulting figures with those obtained by calculation 

 from Mr. Eaton's means, and the result in this case was equally 

 satisfactory with that for temperature already mentioned. 



In conclusion, the Fellows may perhaps be reminded that on 

 June 18th, 1874, one of us (Mr. Scott) read a paper " On the use of 

 an Amsler's Planimeter for obtaining mean values from Photographic 

 Curves," " Proc. Roy. Soc," vol. 22, p. 435. This paper contains a 

 table exhibiting the differences between the means so obtained and 

 those yielded by the hourly values. 



We reproduce this table, appending to it the values obtained from 

 the analyser for the same period. 



It will, of course, be remembered that the mean is the only result 

 which can be got from the planimeter, while it is but a small part of 

 what is yielded by the harmonic analyser ; but a comparison of the 

 figures obtained by the two instruments from the same photographic 

 curves may be interesting, as being got in the case of the planimeter 

 from an instrument in which there is a combined "rolling and slipping" 

 movement, while the movement in the analyser is one of " pure 

 rolling contact." 



2 d 3 



