244 Mr Arago 07i the Thermometrical State 



of the excavations under the Observatory at Paris. They are 

 about 90 feet deep. Thermometrical observations have been 

 made there for a century and a half, and all that seems neces- 

 sary is to examine these observations. But, without going back 

 to the older instruments, for their graduation is not now well 

 known, it must be remarked that a recent discovery has render- 

 ed the solution of the problem not a little difficult. It is now 

 proved, that in the course of time almost all thermometers be- 

 come untrue. The point indicative of the freezing of water 

 mounts upwards along the scale, as if the bulb contracted upon 

 itself. The thermometer will thus then come to indicate 33° or 

 34°, when it ought to indicate 32°, and 35°, instead of 34°, &c. 

 &c. The error is sometimes found to extend to 3°. The nu- 

 merous observations, then, made in les souterrains of the ob- 

 servatory, at a time when it was not known that thermometers 

 must be corrected, are as if they had not been. Two obser- 

 vations, however, have been found, but two only, from which 

 something may be extracted. They go back as far as February 

 1776. They were taken by Messier, with a thiermometer, con- 

 structed under his own inspection, and verified by himself a Jew 

 days before. These two observations, perfectly agreeing, give 

 11°,8 centigrade, (about 53° Fahr.) And in 1826, half a 

 century afterwards, it has been found precisely the same. We 

 shall suppose, now, that in the observations of Messier, on ac- 

 count of the smallness of the scale of his thermometer, there was 

 an uncertainty to the extent of one-twentieth of a degree. These 

 two temperatures of 1776 and 1826, which have appeared to us 

 equal, would then have differed to this extent. But one-twen- 

 tieth (^^) in fifty years is J^ in a hundred, which would yield 

 wily one entire degree qf variation in a thousand years. 



The two epochs which we have compared comprehend be- 

 twixt them a period during which some portions of France have 

 been deprived of their woods to a very great extent. The mean 

 temperature of Paris, however, seems to have undergone no ap- 

 preciable change. 



We have taken the observations of 1826 as the term of com- 

 parison, that we might thus have a round number of fifty years. 

 Had we pushed the comparison farther, brought it down to 1833, 

 we should have found that observation yielded 7 centiemes of a. 



