THE KEW NATIONAL STANDAED OF LEN&TH, AND ITS PEINCIPAL COPIES. 651 
“ I will now say a few words on the question of measuring temperature, leaving the 
measurement of the expansion of the bars aside. 
“ A httle consideration showed me that there was no use in making most accurate 
comparisons, unless I had most accm’ate means of measuring the temperature also ; and 
that I required to know the reading of the thermometer to at least the -^th of a degree 
of Faheexheit. I soon found that there was no reasonable expectation of procuring 
such thermometers, and I proceeded as follows : — 
“ Mr. Simms assigned me a very intelligent and zealous workman, who, after some 
disappointment of external aid, undertook to graduate the thermometers himself. The 
tubes were divided by the dividing engine (probably as well as it could be done) to 180 
equal parts, from somewhere below freezing to about boiling. He then bit-in the divi- 
sions by fluoric acid. 
“ The thermometer with the tube thus prepared comes into my hands. By a column 
of mercury occupying about 90 parts, the whole length is bisected and the error of the 
90 divisions ascertained, and also the error of 30 relative to 120, and that of 150 relative 
to 60. Next with a column of mercury occupying about 60 parts, which is carried in 
three steps from 0 to 180, the errors of 60 and 120 are found, and thus (as above) those 
of 30 and 150 are known. Also by carrying the column of 60 parts backwards and for- 
wards fr’om 90, new measures of the errors of 30 and 150 are obtained ; if these agree 
with the former, the errors of 30 and 150 are well known. Then with a column of 
about 50 parts, which is carried in three steps from 0 to 150 and from 30 to 180, the 
divisions 50, 100, 80, 130 are corrected; and the length of this 50 column being now 
well known, it is used by application to the former divisions to subdivide to ev^ery 10. 
A column of 45 parts will then subdivide all to every 5, and this subdivision is checked 
by the comparison of each 5 with the mean of the adjacent 10. I have learned the 
manipulation of a thermometer so as to break the column myself, and have thus saved 
much trouble. [Mr. Sheepshanks then described the practical method of performing 
this operation.] 
“ The next thing is to settle the boding- and freezing-points (as all the determinations 
already described are referred to degrees 0 and 180, which may be erroneous). There 
is no difliculty as to the freezing-point ; a great number of observations showed me that 
with proper precautions I could rely on it to about r^th of a degree. The boiling- 
point presented some difliculty, until the following method was adopted. The water is 
boiled in a close vessel having apertures for the escape of steam under control, and these 
are regulated till there is just no sensible pressure on a mercurial siphon communicating 
with the interior. [The bulb, as I understand, was surrounded by steam.] The thermo- 
meter may now be considered as completely made. 
“ But there is an anomaly which ofiers some difliculty. Take an old thermometer, 
ascertain its freezing-point, boil the thermometer, ascertain its freezing-point again, 
and you will And it has fallen from one-third to one-fourth of a degree. I found a 
sensible though not a large change by raising the temperature to 100°. I believe 
