170 PROCEEDINGS OP THE AMERICAN ACADEMY 



the results is a value which is essentially identical with the average 

 values from the other two series. The probable error in this last 

 series is larger than in the second, although still very small ; but the 

 difference is due, as the note-books plainly show, to the different con- 

 ditions under which the two series were made. As before stated, the 

 condensation of the balance was perfect, and the apparent weight of 

 the globe did not alter by a tenth of a milligram, even with wide 

 variations of temperature and pressure, so soon as those changes be- 

 came constant. But when the changes of temperature in the balance- 

 room were rapid, currents of air were established in the case, however 

 great care was taken in protecting it, which rendered the apparent 

 weight irregular to the extent of one or two tenths of a milligram ; 

 and the third series was made under less favorable conditions in this 

 respect than the second. This point is illustrated by the following 

 notes of two determinations, which are given in full, in order that all 

 the circumstances connected with the determinations may be seen. 



Series II. Determination 5. 



Weighings of the globe : — Grams. 



Exhausted. June 6th, 6.00 p. m. Tare = 0.1960 



0.2011 

 Filled with Hydrogen. June 7th, 7.20 p. m. Tare = 0.6100 



