1831 .] 
On the Errors of Thermometers. 
87 
VI . — On the Errors of Thermometers, and on a correct Method of 
Graduation. By Lieut. R. Shortreed, Bom. Est. 
The errors of Thermometers are so notorious, that scarcely can any two be found 
to agree in their indications from freezing to boiling, unless they be graduated from 
each other, or from a common standard. A great deal of the discrepancy in the 
results of the different observers, are doubtless to be attributed to this cause alone.—* 
I shall give one instance out of the many which might be produced. — From the ex- 
periments of Mr. Southern and Dr. Ure, on the force of steam at different temper- 
atures, it would appear, that their Thermometers, though agreeing at 212, differed 3® 
at 290 ; Dr. Ure's 290 corresponds almost exactly to Mr. Southern’s 293, and there 
is a gradual increase in the difference of their indications at all the intermediate 
temperatures. — In order to ascertain whose thermometer is in fault, it is very desira- 
ble, that each of them should be examined by the method which I shall have occasion 
presently to propose. It is quite possible that both thermometers may be wrong, 
but both cannot be right. Perhaps I may have occasion, hereafter, to advert to this 
subject. 
When the freezing and boiling points are properly ascertained, the next object is 
the graduation of the scale. The methods of doing this are generally very defective 
in several respects. They may be seen at length, in the article Thermometer, pub- 
lished by the Society for the Diffusion of Useful Knowledge. 
When a thermometer is properly filled, and exhausted, the mercury will flow along 
the tube, when it is reversed and shaken ; and on holding it upright, it will flow again 
into the bulb. On repeating the operation, it maybe observed, that the mercury will 
begin to leave the bulb at the point which was last filled, or the vacuum in the bulb 
will commence from the place where.it previously existed. If the mercury be mada 
to flow along the tube, and at the instant be quickly brought into a vertical position 
and shaken so as to bring the vacuum exactly under the bottom of the stem before 
the bulb becomes full, it will be found, on reversing, that now the mercury will not 
flow from the bulb as formerly, but the whole thread of mercury in the tube, will 
separate from the bulb. This is commonly considered to be an imperfection in the 
instrument, but it is easily remedied, as will presently appear ; and it is by means of 
this thread that I propose to graduate the scales of all thermometers correctly, in- 
dependent of any reference to a common standard or register. When the thread 
separates in this way from the bulb, hold it in a vertical position, and adjust the 
temperature, so that the length of the thread, in the tube, may be nearly, or exactly 
equal, to half the distance between the freezing and boiling points. In this state re- 
verse the instrument, so as to separate the thread from the bulb, and bring the lower 
end to coincide with the freezing point, and carefully mark the termination of the 
upper extremity. Then make the upper end of the thread to coincide with the 
boiling point, and mark, in like manner, the termination of the lower extremity. If 
these two marks coincide, we have got the exact middle point on the scale ; if not, 
the mean of the two is the middle point. It is desirable to repeat the operation, so 
as, if possible, to get both to coincide — and to prove the former trial. On the centi- 
grade scale, freezing being 0°, and boiling 100°, the middle point, so found, is the 
correct 50, and in like manner may be obtained, equally correctly, the points cor- 
responding to 25 and 75. 
We thus obtain three intermediate points, as correctly as the freezing and boiling 
points ; and if the distances of these points be tolerably uniform, the intermediate 
degrees may be obtained in the ordinary way by simple division. 
