562 
Proceedings o] the Royal Society 
6. Researches in Thermometry. 
By Edmund J. Mills, D.Sc., F.R.S. (Communicated by 
Professor Crum Brown.) 
{Abstract.) 
Having had occasion in the course of my work to investigate 
some of the properties of the mercurial thermometer, I have 
obtained a series of results which are comprised in a memoir now 
submitted to the Society. A brief summary of these is given in 
the following abstract. 
1. After describing a simple method of calibrating a thermo- 
meter, I give a detailed proof (following Pierre) that the cali- 
bration unit gradually undergoes a slight diminution in value. In 
the course of five years this may amount, in a new thermometer, 
to as much as *21 per cent. The 0°— 100° interval, therefore, 
requires periodic verification. 
2. When the indicating part of a thermometer has a different 
temperature from the bulb, an ‘‘exposure” correction becomes 
necessary. If y represent the value of this correction, it is 
generally determined from the equation — 
y = *000 1545(T - fj'N ; 
where *0001545 is the difference between the co-efficients of 
cubical expansion of glass and mercury, T is the reading of the 
thermometer, and t is the mean temperature of the exposed 
portion N. Experimental evidence is adduced in the memoir to 
show that the factor of (T-£) N is not a constant quantity, hut a 
linear function of H. The equation thus becomes — 
y-(« + /SN)( T-fJN, 
where a and j8 are constants to he determined from the experi- 
ments. The values of a and (3 are very small, and from about 
1500-2000 eye-observations were required to determine them, 
according to the instrument employed. 
3. The gradual ascent of the zero with time can he expressed 
by a logarithmic curve having two terms, viz. — 
y = Aa* + B j3 x , 
