OF ARTS AND SCIENCES. 



59 



and it was evident that the insignificant differences remaininsf arose 

 from the circumstance that the thermometer was not nearly so sensi- 

 tive as the air in the balance-case, following the changes of tempera- 

 ture of" the air after a considerable lapse of time. It was curious to 

 notice the slight increase of weight, caused by the radiation of the 

 body while weighing, followed after some time by a rise of the very 

 sensitive thermometer employed ; and this effect was obtained with 

 a displacement of only about 75 cubic centimeters of air. 



TABLE OF WEIGHTS. 

 Second Series. 



Note. — In combining only the extreme weights, we must obviously take care that neither of 

 them is seriously affected by any accidental errors ; and a more certain value of the constant 

 would be obtained by combining all the observations after well-known methods. This compli- 

 cation, however, is seldom necessary ; as such errors would render the final result irregular, and 

 lead to a rediscussion of the observations. 



The limits in the accuracy of the method here described are obvi- 

 ous ; but it will be noticed that the accuracy of the method is exactly 

 proportional to the requirements. The greater the volume of the 

 load, and hence the greater the effect of buoyancy, the more accu- 

 rately can the " constant " be found, by which the correction required 

 in any case can be calculated ; and, as the above examples abundantly 



