DETERMINATION OF ALTITUDES BY BAROMETER. 



113 



and thus broke the barometer in a way very difficult to repair. It was successfully accomplished, 

 however, by putting a little of Husband s adhesive plaster on both sides of the crack, and 

 then covering it with sealing wax dissolved in alcohol, to protect it from the iir. 



INSTRUMENTAL ERRORS. 



In order to eliminate the effect of capillary attraction, of minute bubbles of air which cannot 

 be entirely excluded from a tube unprovided with Daniell s protective ring, and of other causes 

 of instrumental error, the scales of all the barometers were so adjusted by the maker that the 

 instruments agreed precisely with Smithsonian standard on leaving New York. On reaching 

 Fort Reading, from a mean of over two hundred observations, taken with great care by the 

 gentlemen of the party, Nos, 1060 and 1061 were found to agree exactly with each other, while 

 both Nos. 1068 and 1089 differed slightly from them. It was assumed that the two former had 

 remained unchanged ; and corrections to make each of the others agree with them were deduced 

 from the above mentioned observations, after the temperature of the mercury had been reduced 

 to 32 Fahrenheit. Subsequently, whenever a barometer was broken and re-filled, a similar 

 correction was deduced. The following table exhibits these corrections. 



Barometer No. 1060. 



Date. 



Inches. 



July 10 to September 18 .000 



September 19 to September 2S -f .022 



September 30 to October 15 i -(- .023 



October 1C to October 29 -|- .042 



Subsequent to October 29 -}- .191 



Barometer No. 1061. 



Date. Inches. 



July 10 to August 5 .000 



Irreparably broken on August 5 



I 



Barometer No. 1068. 



Date. Inches. 



During whole survey .009 



Barometer No. 1089. 



Date. Inches. 



July 10 to August 5 -f- .015 



August 6 to September 4 + . 044 



Subsequent to September 4 + 035 



15 X 



