ON THE MECHANICAL EQUIVALENT OP HEAT. 
381 
Freezing-point Thermometers. 
22. Two similar thermometers were obtained, one only of which was ever used 
during the experiments. This was a chemical thermometer, bearing the laboratory 
mark 2Q, with a :^-inch diameter stem having its scale very plainly etched in black 
lines on the glass. The length was 11^ inches over all, the bulb being l-g inches 
long, and then at a distance of 2^ inches from the top of the bulb the graduations 
began. The scale extended from 30° to 45° Fahr., 6f inches of the stem being 
occupied by the 15° mentioned. Each degree was divided into tenths, and it was 
easy to estimate to the hundredth of a degree. 
The index error of this thermometer was repeatedly checked during the whole 
period occupied by the research by being immersed in a mixture of pounded ice and 
water. 
The table appended gives the corrections and the dates on which tests were 
made :— 
Date. 
Reading. 
Correction. 
5th December, 1895 .... 
31-7 
+ 0-3 
20th December, 1895 .... 
31-71 
+ 0-29 
9th January, 1896 .... 
31-67 
+ 0-33 
17th January, 1896 .... 
31-67 
+ 0-33 
31st January, 1896 .... 
31-57 
+ 0-43 
5th February, 1896 .... 
32-48 
- 0-48 
20th February, 1896 .... 
32-46 
- 0-46 
16th March, 1896 . 
32-46 
- 0-46 
21st April, 1896 . . 
32-47 
- 0-47 
25th June, 1896. 
32-47 
- 0-47 
7th July, 1896 . 
32-52 
- 0-52 
Before making the test on January 31st the hot water from the brake backed up 
round this thermometer, so that the sudden alteration in the reading is accounted for 
to some extent. 
Also up to this time part of the mercury had remained stuck in the upper bulb, 
but Dr. Harker, of the Physical Department, now succeeded in bringing the 
separated mercury down into contact with the column below\ 
By permission of Dr. Schuster the scale of this thermometer was compared by 
Dr. Darker on the 27th April, 1896, with a standardised thermometer (Baudin, 
No. 12,771) in his possession between the points 32° and 35° Fahr. 
This comparison showed that the correction of — 0’47 as obtained on April 21st 
was correct between 33° and 34°, which was the part of the scale used in most of 
the experiments up to that date. 
At 35°, however, the correction increased to — 0’5, and consequently in the later 
experiments, when the temperature of supply in the heavy trials approached this 
