DISCUSSION OF BAROMETRIC OBSERVATIONS, 



475 



of the fluctuation of mean range. But there is an additional peculiarity 

 of tliis fluctuation which appears to have nothing in common with that 

 of the solar spots except equality of period. 



I'rom each year of minimum sun-spots and maximum range to the 

 year of the opposite phases the progress is always uninterrupted, but 

 the rise of mean range from the latter x^oint to its next maximum on the 

 year of the sun-spot minimum is checked in every instance nearly mid- 

 way, or when the sun-spot number is about 40, by a decline lasting in 

 each of the four cases through two years, after which the increase of 

 range is resumed. 



The highest corrected barometer-reading since 1853 is 30.G87 on Feb- 

 ruary 12, 1857, and the lowest is 28.509 on i^ovember 18, 1873. Their 

 difference, the range in twenty-five years, is 2,178 inches. 



Of the appended diagrams, the first (divided into five sections, and 

 marked Parts 1 to 5, forming continuous curves for the year) shows the 

 barometric i)ressure for each day of the year, as given numerically in 

 Tables I, II, and IV. The full (or solid) line exhibits the daily values of 

 pressure directly from observations (Table II), while the fine dotted 

 curve shows the normal values computed from the periodic function 

 (Table I), and the intermediate broken (or coarse dotted) curve is taken 

 from the values given in Table IV. Vertical lines are inserted at inter- 

 vals, having the middle of their length placed at the height of the 

 annual mean (29.719), and showing by their extent upward and down- 

 ward the probable error of Table II for the date. The i^robable error 

 of Table IV is also indicated by corresponding marks upon the vertical 

 lines. 



Probable eirors of Tables II and IV, for each month. 



The concluding diagram exhibits graphically the connection between 

 the fluctuations of barometric range and of sun-spot amount, as given 

 in Table VII. 



