VOL. LXXI.] PHILOSOPHICAL TRANSACTIONS. l63 



effected with more certainty and precision. After various experiments I hit on 

 the following method, which seems both new and capable of great accuracy, 

 though in this I may be deceived. 



First, having blown a ball to a capillary tube, such as are commonly used for 

 thermometers, I weighed it, and found that this empty thermometer was equal 

 to 79-25 grains. This empty glass, previous to its being weighed, was rendered 

 as perfectly clean as possible, which is a necessary precaution in this experiment, 

 which depends on a very great accuracy of weight. I then introduced some 

 mercury into the stem of this thermometer, taking care that none of it entered 

 the ball, and, by adapting a scale of inches to the tube, observed that 4.3 inches 

 length of the tube was filled with the mercury. The thermometer was now 

 weighed again, and from this weight, the weight of the glass found before being 

 subtracted, the remainder, viz. 0.24 gr. showed the weight of so much quick- 

 silver as filled 4.3 inches of the tube. Now the ball of the thermometer, and 

 also part of the tube, were entirely filled with quicksilver; then, to find out the 

 weight of the mercury contained in it, the thermometer was weighed for the last 

 time, and from this weight the weight of the glass being subtracted, the re- 

 mainder, viz. 32.05 gr. showed the weight of the whole quantity of quicksilver 

 contained in the thermometer. 



By comparison with a graduated thermometer in hot and cold water, I made 

 a scale to the new thermometer according to Fahrenheit's, and by applying a 

 scale of inches found, that the length of 20° in this scale was equal to 1.33 

 inch. But 0.24 gr. was the weight of so much mercury as filled 4.3 inches 

 length of the tube; therefore, by the rule of proportion it will be found, that 

 the weight of so much quicksilver as fills J .33 inch of the tube, viz. the length 

 of 20°, is equal to 0.0742 gr. nearly, and that the weight of so much quicksilver 

 as fills the length of the tube that is equivalent to 1°, is equal to 0.00371 gr. 

 Now it is clear, that the weight of the whole quantity of quicksilver contained 

 in the thermometer, is to the weight of so much quicksilver as fills the length 

 of 1° in the tube, as the bulk of the whole quantity of quicksilver in a given 

 degree of heat, to the increase of bulk that the same whole quantity of quick- 

 silver acquires when heated of but 1°, viz. 32.05 gr. is to 0.00371 gr. as 1 is to 

 0.001 1 + ; so that by this experiment it appears, that 1° of Fahrenheit's ther- 

 mometer increases the bulk of mercury not above T i) l' 00o parts. In this process 

 a small deviation from mathematical exactness is occasioned by the small difference 

 of weight between the quicksilver of the tube when first weighed and when 11 is 

 afterwards heated to 1°; but by an easy calculation it will be found, that this dif- 

 ference is so exceedingly small as not to be perceived by our exactest weighing 

 and measuring instruments. 



For clearness sake I shall subjoin ihe calculation of the above related experi- 



y 2 



