M E T E R O L O G Y. 



155 



Meteorolo- the original paper, in which this opinion is said to be 



fj- advanced, we cannot give the solution of the qm 

 s ~<' ' ; Humboldt's own words ; but if the abridgment of it 

 Afratui. . ^ in Tnon)son - s Annals can be depended on, it ap- 

 w^jSL. pears that the principle on which that opinion rests is 

 merely a gratuitous assumption. The words of the 

 abriilger are these: " By comparing a great ninntwr 

 of observations made between 4&> and 4b', X. latitude, 

 we find that at the hour of sunset the temperature is 

 nearly the mean of that at tunri e and tn/o hour* 

 a/ter noon.'" The amount therefore of Humboldt's dis- 

 covery is simply this: Thai the temperature at sunset 

 agrees i-.ore nearly with the mean of sunrise and two 

 hour* alter noon, than it does with the mean of the max- 

 imum and minimum observed during the day and night; 

 -. does not necessarily follow that it agrees with 

 erature of the day. It setm< doubt- 

 : himboldt ever intended to draw any 

 such concluion. In the paper already alluded to, the 

 reason assigned for making observation* at sunrise and 

 two hour* after noon, is, that the temperature* at these 

 hours are considered as indicating the maximum and 

 minimum. It is al*> stated, that if, beside* the msxi- 

 .!;! minimum, a middle observation be taken, it 

 should be at least four or five hours from either of the 

 others; but that, upon the whole, the to oiufTaiujMf 

 rme Ifm/ffraluifi ill give uc more correct re- 

 me* therefore that llum- 

 . lately appeals, as the standard by which to 

 pcrature of any other hour ap- 

 proaches to the true mean, and if he has recommend- 

 ed any fixed period for making a single observation, it 

 is rather with a view, we suspect, of abridging the 

 labour, ti.an of affording a more correct result. Such 

 an abridgment is certainly desirable, but we are dis- 

 posed to doubt whether it is at present attainable in 

 the case of meteorological observation*. The science 

 is still in a condition that requires a copious induction 

 of facts ; and there i* reason therefore to apprehend, 

 that in proportion a* we diminish the number of ob- 

 servations, we sacrifice accuracy to simplicity, and sub- 

 stitute theory for experiment. 



Mean of ^ ur readtra will perceive, from the*e remark.s, that 

 ta* daily we are ii.c'i I i consider the mean of the <Uily ex- 



we are inclined to consider the 



tremes, ai approaching nearer thnn that of any other 

 aypraxtua observation*, to the true mean temperature of any 

 ib Ira* given place; and the ground* of this opinion appear 

 '" to us exceedingly simple and o!n ,-. Were the ac- 

 tual temperature observed st very short intervals, say 

 every half hour during the day and night, through the 

 whole year, the mean of all the observations would 

 undoubtedly give a very near approximation to the 

 true annual mean ; but it would obviously be still 

 nearer if, instead of these, the extremes of every hour 

 were marked, and half their sum taken as the mean of 

 the hour. The former method, indeed, would be equal- 

 ly accurate, did the temperature of the atmosphere 

 suffer no interruption in it- progress from one extreme 

 to tl. it liable as it is to frequent fluctuations, 



the i. !>le method of approximating to a cor- 



rect estimate of these, is to note the extremes at short 

 interval*. The same reasoning will apply, in some 

 degree, to interval* of a greater length, though it may 

 naturally be supposed, that, in proportion a* these are 

 extended, the result will be less accurate. We have 

 vever, from actu J observation, that the dif- 

 ference between the means nf the extreiin- tempera- 

 ture*. shorter and longer intervals, is much 

 less than at first sight we should have been led to ex- 

 pect, as will appear by the following results, deduced 

 from our observations for 1817 and 1818. 



MIT, 



Mean of the 2 annual extremes . . 45.5 

 Mean of the Iv! monthly extremes . . 45.4 

 Mean of the 36' extremes (breach 10 days IV J. 

 Mean of the 365 daily extremes . . -Hi.1 



These results, we think, furnish satisfactory evidence Adranta- 

 th at the mean temperature derived from observing the *" d n f "' 

 daily extremes, cannot be far from coinciding with the j[ : _"^ i _ 

 true mean of the whole year ; but even granting that ,,.*" 

 it were otherwise, there are still arguments enough to 

 trate that these extremes outfit to be recorded, 

 in preference to any other observations. In determin- 

 ing the law which regulates the distribution of heat 



lie globe, we must be guided entirely by experi- 

 ment. Calculations of the quantity of heat which is 

 communicated to the earth in a given time, by the di- 

 rect action of the sun's rays, or of the effects produced 

 by the constant interchange of warmer and colder por- 

 tion* of the atmosphere, between one region and ano- 



nuy be sufficiently amusing ; but we shall look 

 in vain to such calculations for fie ui-covery of a law, 



. as we have every reason to believe, is affected 

 by so many disturlnng cause*, and modified by such a 

 instances). The combined effects of all 

 these can be discovered only by collecting the results 

 of observations in various ehaaUons, and under differ- 

 ent latitude*. Nor u it enough that these observations 

 have been carefully and regularly made. Before they 

 can be employed as the foundation of any theory, it 

 must be ascertained that they were made under similar 

 circumstance*). To compare the mean temperatures of 

 two different places, which have been deduced from 

 observation* made at different hours, is to compare 

 quantities that do not admit of comparison, and which, 

 tor any thing we know, may bear very different rela- 

 tion* to the true mean* of these places. Now, ihe 

 great advantage of recording, in nil c:ie, the maxi- 

 mum and minimum of the day and night, v.-ou'd be, 

 that every meteorological register wmdd .[>cak the 

 same language, r.nd put us in pos-esvlon of the same 

 physical la ' if they d:<( not give us the true 



mean temperatures of the respective places where they 

 are kept, (though, as we have already seen, their re- 

 sults would approximate to these quantities,) still the 

 term mean Umpernturc, which at present is used by 

 every observer to denote the annual average of his ob- 

 servations, at whatever hours they may .have been 

 made, would have a definite anil distinct meaning, ex- 

 pressing, in all caoes, the middle point of the thermo- 

 me'rical range during certain period*. Till sonic such 

 uniformity be established among meteorologists, the 

 multiplication of registers will serve but little to ex- 

 tend the boundaries of this branch of science. 



the reasons now stated, we are not di -posed to Mayer's 

 place so much confidence ns many have done, in the formula 

 theorem given by Mayer, for determining the mean nnoi be 

 temperature of any given latitude nt the level of the founded o 

 sea. We do not indeed quarrel with it because it is " 

 avowedly empirical, for it is only from experiment 

 that any satisfactory rule on the subject can be deduc- 

 ed ; but we entertain very considerable doubts as to 

 the accuracy of the observations on which it is found- 

 ed. We have already seen, that, from the diver-: 

 opinion that still exists respecting the best lime for ob- 

 serving the temperature of the atmosphere, many of 

 our meteorological registers are comparatively of little 

 use ; and we have no reason to suppose, that, previous 

 to the middle of the last century, they were kept on a 

 more accurate principle. The theorem, however, is 

 valuable, as affording, by a very simple process, a gc- 



