74 M. Bouvard on the Diurnal 



According to this table the greatest heights of the barome- 

 ter during the year take place in January \ and the smallest 

 in April and October. The excess of the maximum of the 

 height above the minimum is 3.39 millimetres, a quantity 

 which indicates that the uncertainty of the mean absolute height 

 of the barometer at Paris ought to be about 0.15 of a millime- 

 tre more or less. 



This table also shows that the extent of the barometric pe- 

 riod is not the saihe for each month. Its variation does not 

 appear to have any connection with that of the height of the 

 barometer, for this period preserves the same value, whilst the 

 mercury passes from its greatest to its smallest height. But in 

 examining theVesults for the 132 separate months, we find, as 

 M. Laplace had already recognized from calculations which I 

 had communicated to him some years ago, that the mean re- 

 sult of the diurnal variations from 9 h a. m. to 3 h p. m. in No- 

 vember, December, and January, was in each year regularly 

 less than in the three following months of February, March, 

 and April. The mean variation indeed of eleven years was 

 m .557 for the three first months, and m .940 for the three 

 last. The mean of the first six months was m .748, nearly 

 equal to the mean daily variation of eleven complete years. 

 The other six months present nothing similar ; but we find 

 m .752 to be the mean daily variation of the months of May, 

 June, and July, and m .802 for that of August, September, 

 and October, the mean of these six last months being m .777. 

 There is, therefore, some annual cause yet unknown which in- 

 creases in February, March, and April, and diminishes in No- 

 vember, December, and January, while it preserves an in- 

 termediate value during the other six months of the year. 

 This phenomenon is well established. It cannot be the effect 

 of chance ; and it will be interesting to discover the principle 

 on which it depends. 



It would be in vain to seek in the daily variations from 3 h 

 p. m. to 9 h p. m., a phenomenon analogous to that which we have 

 mentioned as existing in the period from 9 h a. m. to 3 h p. m. 

 The last column of the table shows that the value of this pe- 

 riod scarcely changes m .3 in the year, and that its oscillations 

 follow a regular law. 



