332 
BULLETIN OF THE UNITED STATES FISH COMMISSION. 
COMPARISON OF THERMOMETERS. 
We determine tlie error of a thermometer at the temperature of the freezing point 
by completely covering the bulb and that portion of the stem containing mercury with 
clean ice shaved u}> into line fragments by a Jack-plane. This test must be made in a 
jdace where the air temperature at the time is above the freezing point, so that the 
ice will melt steadily. After being exposed to this condition for five or six minutes a 
reading of the thermometer shows whether or not it is correctly graduated and the 
amount of any error that may exist. 
For any other temperature than freezing, excejit the boiling temperature, the 
thermometer to be tested and a standard are placed in a bath of water, if the 
temperature is above freezing, or of alcohol if below freezing. The liquid being 
tlioi'oughly stirred so as to render the temperature uniform throughout, a quickly 
made reading of the two thermometers shows the amount of error in tlie graduations 
of the one under test. 
When a large number of instruments are to be compared with standards, as is the 
case at the Weather Bureau, the thermometers are collected together in bunches of 
12 each. Only the glass tubes are placed in the bunches, the metal backs of the 
thermometers being removed during comparison, not only to avoid bidkiiiess but to 
])reveut the injurious effects and slight corrosion of the metal caused by the water 
and alcohol baths. The bunches are formed upon flat metal frames adapted to 
receive six thermometers on each face, front and back, the thermometers being held 
by rubber bauds. 
For the freezing-point test a small wooden box 5 inches wide, 4 inches deep, and 
18 inches long is employed. To permit the water from the melting ice to escape, the 
bottom is pierced with irregularly distributed holes 1 inch in diameter and about 24 
inches apart. This box is nearly filled with clean, shaved ice, bunch after bunch of 
thermometers are placed therein and new ice added and packed closely ai ouud each 
thermometer and jiiled up above the sides of the box until the bunches are wholly 
covered, being supported by the ice. 
The readings of the thermometers are then made and recorded, the ice being 
scraped away so as to expose the stems in the vicinity of the ends of the mercurial 
columns. A small eye-lens is always employed to assist the vision and to insure that 
the line of .sight is exactly at right angles to the stem of the thermometer. 
This reading glass consists of a small lens having a focal length of about 1-J 
inches. It is set in a brass tube 4 inch in diameter and 3 inches long. The lens is 
about inches from one end of the tube and the opposite end is closed by a metal 
cap having a small hole in the center, about one sixteenth of an inch in diameter. In 
reading a thermometer the open end of the brass tube is set squarely against the 
glass stem while the eye sights through the small hole in the cap. Tlie eye estimates 
fractions of a degree to the nearest tenth, or, if the graduations ai'e halves or fifths of 
degree.s, then the estimation is to the nearest hundredth of a degree. 
The tests at the freezing temperature are followed by tests at tem])eratures 42°, 
52°, and so on, for every ten degrc'es up to 102° or 112°. The apparatus employed 
consists of two cylindrical co])per cans, one within the other. The outer can is 12 
inches in diameter, the inner 8.5, and each is 13 inches high. Three feet upon the 
bottom of the inner can make a space of three-(piarters of an incli Ijetween the bottoms 
of the two vessels. The inside vessel is fitted with a light metal frame in the form of 
