529 



These results will be obvious at a glance ; but a few remarks 

 upon the instruments employed, and upon the degree of reliance to 

 be placed upon them, may not be uninteresting. 



The dry- and wet-bulb thermometers (for which we were indebted 

 to the kindness of Prof. Cherrirnan, Director of the Magnetic Ob- 

 servatory, Toronto) were made by Negretti and Zambra, and their 

 index errors were ascertained, above 32 by Mr. Glaisher, and below 

 32 by ourselves, by comparison with a Kew standard. The divi- 

 sions upon these thermometers were too small to read 0'l with 

 great accuracy ; and in discussing our observations at low tempera- 

 tures, we were in consequence obliged to reject such as would, with 

 an error of 0' 1 in the reading, introduce a considerable error into 

 the factor. 



You will observe that the table does not extend below 16, 

 although we have repeatedly every winter the mercury below 20, 

 and occasionally below 30. The only thermometer, however, 

 which we could trust as a wet-bulb in investigations so delicate was 

 not graduated below 16. 



For obtaining the dew-point by direct observation, we used the 

 condensing hygrometer invented by M. Regnault. 



We obtained dew with this beautiful instrument at all tempera- 

 tures (limited only by the graduation of the thermometer 35), 

 the only requisites when the thermometer is very low being time 

 and pure ether*. I can testify from experience that this hygro- 

 meter obviates all the disadvantages of Daniell's, which M. Regnault 

 enumerates in his hygrometrical researches. 



In order to show the reliance that may be placed upon our results, 

 we have put opposite each factor in the table the probable error and 

 measure of precision of the single data (from which the factor (f) was 

 derived), and also the probable error, measure of precision, and limits 

 of certainty of the adopted factor. The nomenclature and notation 

 are thus employed by Encke in his Memoir on the Method of 

 Least Squares. 



The measure of precision (Ji), as was indeed to have been ex- 

 pected, decreases with the temperature. This fact is not however 

 of so much importance as might at first appear. 



* The ether we employed below 20 was the first that passed over, resulting 

 from the distillation of washed ether with quicklime. 



