32 REPORT — 1841. 



produced on the heights when the barometei' is at any point below 29'4< 

 inches or 29'5 inches, is always greater than the proportion for the greater 

 heights of the barometer. I imagine this arises from the effect of wind, 

 which generally follows a great depression of the barometer, and generally, 

 with us, comes from the S.W. ; so that an additional elevating cause comes 

 into operation. This, however, is mere conjecture." 



I then requested that, instead of arranging the observed heights according 

 to the barometer, he would correct the observed heights for lunar and solar 

 parallax and declination, and investigate the effect of atmospheric pressure 

 on the residues ; still comparing the contemporaneous and the anterior epochs. 

 The following was the result. 



"March 17, 1841. 



"I send you the results of comparisons of the residues of height for 1834, 

 1835 and 1836, with the state of the barometer at different epochs. The 

 heights Avere calculated carefully by numbers, using what I consider to be 

 my best corrections for lunar and solar parallax and declination, and em- 

 ploying the same corrections in each of the three years. The only correc- 

 tion omitted was that for the diurnal inequality. The residues for 1834< 

 were compared with the barometer contemporaneous, twelve hours anterior, 

 twenty-four hours anterior, twenty-four hours posterior, and the extreme 

 groups, with barometer, thirty-six hours anterior ; in order to find what pro- 

 gressive changes of form the curves would thus be made to assume. The 

 mean correction for one inch difference in height of barometer having been 

 obtained, the proportional correction was applied to each observed height of 

 high Avater, and the mean of all the errors (remaining after the barometrical 

 correction) then taken for the whole year. In every instance the contempo- 

 raneous barometer gives the best correction. Thus in 1834 the mean error 

 remaining, after applying the barometrical correction, is 



5"817 inches contemporaneous barometer. 

 6'085 inches barometer twelve hours anterior, 

 6*221 inches barometer twenty-four hours anterior, 

 6*248 inches barometer twenty-four hours posterior. 



"These two latter epochs, the one a7iterior the oihev posterior, producing 

 nearly equal errors, seem to show (like equal altitudes) that the truth lies 

 midway between them, 



" In like manner, the mean residual error for 1 835 is 



5'277 inches, barometer contemporaneous, 

 5"421 inches, barometer twelve hours anterior, 

 5*706 inches, barometer twenty-four hours anterior ; 



and for 1836 is 



6*450 inches, barometer contemporaneous, 



6*535 inches, barometer twenty-four hours anterior. 



" The introduction of the correction for the contemporaneous barometer re- 

 duces the mean error, previously remaining, about one-fourth, being as 1 : 

 0*753, for the whole of the year 1834 ; and as 1 : 0*705 for the year 1835. 



" The mean effect on the tide corresponding to a change of one inch in the 

 mercurial column was carefully obtained, by taking into account the number 

 of observations in each parcel, so as to get the true average. The con- 

 temporaneous barometer gives, in every instance (as shown in the diagrams), 

 the greatest result : and in this case also equal differences from the maximum 

 attend the anterior and posterior epochs for 1834 — viz. 11 inches tide (instead 

 of 13*4 inches) to one inch of mercury. 



I 



