328 PROCEEDINGS OF THE AMERICAN ACADEMY 



ature by the following arrangement, which has been found to be mod- 

 erately successful. Sheets of galvanized iron three feet in width, after 

 having been fastened at the edges by soldering, and riveted together at 

 intervals of twelve inches, were fastened to the inner walls of the room 

 in such a manner that there is a gradual decline in their height above 

 the floor. They are connected together by rubber tubing. In the 

 opposite room a tank is arranged for either hot or cold water, which is 

 connected with the reservoir at the highest point. At the lowest point 

 there is an outlet through the walls of the room. Either hot or cold 

 water entering this reservoir of narrow section slowly percolates through 

 the enclosed space, and flows off through the outlet, maintaining a nearly 

 constant supply of either hot or cold air within the room. The reser- 

 voir holds about five gallons, and the amount of surface exposed on one 

 side is about 100 square feet. 



This comparing-room has one fault in construction which has given 

 considerable trouble, and which it has been found impossible to remedy 

 entirely. In order to be able to raise the comparator as high as the 

 window, it was found necessary to give the piers a height of four feet. 

 The head of the observer is therefore very near the ceiling. This dif- 

 ficulty has been partly remedied by a large trap-door in the ceiling, 

 which allows the heat developed by the presence of the observer to 

 enter the space between the partitions. 



Description of the Thermometers employed in the 



Comparisons. 



Throughout the entire series of observations the Yale College stan- 

 dard has been adopted. In the earlier part of the work a Fahrenheit 

 thermometer graduated to fifths of deiirees was used. It is desiirnated 

 0. It was purchased of a dealer in New York at a low price, but it 

 was found to be an exceptionally fine instrument. Its error was de- 

 termined by Dr. Waldo, by comparison with the Yale standard. It 

 was afterwards carefully compared with Cassella No. 3235, which had 

 been rigorously compared with the air thermometer of Professor Row- 

 land by Mr. S. W. Holraan of the Massachusetts Institute of Tech- 

 nology. The independent comparisons of Dr. Waldo and of Mr. Hol- 

 man agree in giving the same tenth of a degree for every point 

 compared. The corrections adopted are given below. 



