180 Humboldt on Isothermal Lines, and the [March, 



authors who have added to our knowledge on the temperature 

 of the different parts of the earth, we must not omit to mention 

 Cotte, who has collected a great number of documents ; but 

 they are not reduced into any kind of method, and are not all of 

 equal authority. 



These considerations lead us to the conclusion that, in inves- 

 tigating the distribution of temperature, it is important to distin- 

 guish between the results which are deduced from observation, 

 and those which are derived from theory. We must collect all 

 the authentic facts that can be obtained ; and, after arranging 

 them into regular order, we must submit them to what may be 

 termed empirical laws. After having correctly ascertained the 

 mean temperature of certain places, we may trace on the globe 

 isothermal lines, which thus exhibit to the eye the relation of 

 these places to each other. In determining the mean temperature 

 of a particular spot, the old plan was to take the maximum and 

 minimum temperature of the year, and to consider the middle 

 number between them as the mean temperature ; but this plan 

 is obviously incorrect. De la Hire seems to have been the first 

 who attempted to pursue a different method, and one founded on 

 more just principles : observing the uniformity of the tempera- 

 ture of the vaults that are attached to the observatory at Paris, 

 he proposed as a general fact, that the temperature of vaults was 

 the mean temperature of the climate.* The other method, that 

 of the maxima and minima, continued, however, to be generally 

 adopted ; and, by multiplying the number of the observations, it 

 was rendered more correct, but still liable to inaccuracy. Some 

 of the latest observers have noted the thermometer three 

 times each day, and then taken the mean of these as denoting 

 the mean temperature of the day ; others have adopted the 

 plan of observing the thermometer at two periods in the day, 

 which are considered as indicating the maximum and minimum — 

 sun-rise, and two hours after noon ; while others, again, have 

 satisfied themselves with observing the temperature at one 

 period only, which has been found, by previous experience, to 

 denote the mean temperature. 



By comparing a great number of observations made between 

 46° and 48° N. latitude, we find that at the hour of sun-set the 

 temperature is very nearly the mean of that at sun-rise and two 

 hours after noon. , When, besides noting the maximum and 

 minimum, we take a middle observation, we shall fall into an 

 error, if we simply divide the sum of the observations by three, 

 without attending to the duration of the particular temperatures, 

 and the place which the middle observation occupies between 

 the extreme tenns of the series. The middle observation should 

 be at least four or five hours from either of the others ; but, upon 

 the whole, the two observations of the extreme temperatures- 



* Mem. Acad. Sciences, 1719, p. i 



