224 DISCUSSION OF THE DESICCATION QUESTION. 



Accurate instrumental observations do not go back for as much as one 

 hundred years. A few facts may be given in support of this assertion ; and 

 in the first place, in regard to temperature records. . - 



A really serviceable thermometer — that is, a thermometer Avhich could 

 be called a scientific instrument — did not exist until Fahrenheit invented 

 his, in 1714. By fixing two points of the scale, he for the first time deter- 

 mined the value of a degree, and made comparisons of temperature at difier- 

 ent times and places possible. But it was not until long after Fahrenheit's 

 day that regular observations began to be taken. In Paris there has been 

 a continuous series from 1763 on, the records taken jorevious to that being 

 more or less fragmentary. In Stockholm regular observations were begun 

 in 1758; in London in 1775. 



The existence of a very serious source of error in thermometrical obser- 

 vations was not noticed until 1817, when Arago called attention to the change 

 of the zero-point of the thermometer used in the subterranean apartments 

 of the Paris observatory.* A few years later, in 1822, Bellani, a Milanese 

 observer, recognized the fact that all thermometers are liable to a change 

 exactly such as would result if the bulb began to grow smaller soon after it 

 was blown, and continued to do so for a long time. Of course observations 

 made previous to this discovery, and, indeed, all observations made without 

 special examinations, from time to time, of the accuracy of the zero-point are 

 of no value for use in any such inquiry as that now before us. 



Arago, in his investigation of the question whether there had been any 

 change in the mean temperature at Paris, as indicated by the thermom- 

 eters observed at a depth of twenty-eight meters below the surface, thought 

 that he could go back as far as the year 1776 for comparison, one obser- 

 vation taken in that year having been made with an instrument recently 

 verified.! Compared with the result obtained with the thermometer, the 

 temperature half a century later seemed to have undergone no perceptible 

 chant'-e, althouo;h the scale of the thermometer observed in 1776 was too 

 small to admit of accurate reading. Arago remarks that there might have 

 been a difference as great as one-twentieth of a centigrade degree, and a 

 little farther on seems to assume that there had been such a change, for he 

 goes on to say : " A twentieth of a degree [centigrade] in fifty years that is 



* CEuvres Je F. Arago. Notices Scieutifiques, Tome V. p. 642. 

 + By Messier. 1. c, p. 645. 



