OF THE ATMOSPHERIC TEMPERATURE. m 



and the hours T, 2, 9 were, in consequence, adopted for the system of meteorological 

 observations at the military posts of the United States, organized in 1819 under the 

 direction of the surgeon-general of the United States Army. Although these hours 

 were at one time abandoned (between 1841 and 1854, when the epochs a little 

 before sunrise, 3 and 9 P.M. were substituted), they were re-established in 1855, 

 mainly through the exertions of Dr. Coolidge, U. S. A. The convenience and 

 satisfactory character of the results of these hours, also led to their adoption in the 

 meteorological observations undertaken conjointly by the United States Patent 

 Office and the Smithsonian Institution in 1854, and they have since been adhered 

 to by the latter Institution. The recognition of the fact that the results by the 

 three hours 7, 2, 9 can be greatly improved by taking one-fovirth of the ordinates 

 at 7, 2, and twice 9 in the place of one-third of the ordinates at 7, 2, 9, appears 

 also to be due to Dr. Dewey. 



From the present collection of results it appears that the homonymous hours, 

 10, 10, give differences of less than J^ 0°.5 in the annual mean, that the triplets, 

 6, 2, 9, and equidistant hours, 6, 2, 10, are of nearly equal value, and but slightly 

 superior to the preceding pair of hours, the former combination producing a higher, 

 the latter a lower mean than the true value of twenty-four equidistant observations, 

 but deviating less than 0°.4. The combination 7, 2, 9, produces a result nearly 

 0°.5 in excess, whereas the modification 7, 2, 9 {bis) diminishes this difference to 

 nearly 0°.l with a change of signs for different stations. The four-hour combina- 

 tion 3, 9, 3, 9, adopted by the Royal Society, is the best of all, being generally less 

 than 0°.l above the true daily mean. In the following table of differences from 

 the daily mean, of the average temperature observed at 7, 2, 9, the sign -|- indicates 

 an excess, the sign — a defect of the latter average. The first line for each station 

 answers to the combination i (7, 2, 9), the second to the modification | [7, 2, 9 (bis)]. 



