Royal Society. 311 



The thermometers " Kew No. 4 " and " Kew No. 14," were 

 graduated on the stems by myself with arbitrary scales : the bulb of 

 No. 4 is spherical, and is about f inch diameter ; that of No. 14 is 

 cylindrical, f inch long and ^ inch diameter, and very sensitive. 

 " Fastre No. 231 (Regnault) " is a standard by Fastre of Paris, also 

 graduated on the stem with an arbitrary scale according to Reg- 

 nault's process. This instrument was examined and approved by 

 M. Regnault ; the determination by him of the scale coefficient 

 agreed closely with that afterwards made at Kew. The bulb is 

 cylindrical, about \\ inch long and \ inch diameter. "Barrow, 

 E.I.C., S 7, No. 4," is one of a number of thermometers made for 

 the East India Company and sent to Kew for examination. Its scale 

 is of brass divided to degrees. " Newman (Makerstoun)" is the in- 

 strument which was supplied to the Makerstoun Observatory as a 

 standard, and to whose indications the results of the temperature 

 observations made there since 1841 have been " corrected." It was, 

 at my suggestion, sent to Kew by Sir Thomas Brisbane for compa- 

 rison with our standards. " Troughton and Simms (Royal Society) " 

 is a standard belonging to the Royal Society. As its scale extends 

 to above 212, its boiling-point was examined in the same apparatus 

 employed for the Kew standards, its brass scale remaining attached 

 to the tube. It was found to read 212°'7 when the barometer, re- 

 duced to 32°, stood at 30-136 inches. 



The errors of a thermometer which has been already carefully 

 examined between 32° and about 100°, may be obtained with con- 

 siderable accuracy for temperatures below 32°, without using a free- 

 zing mixture, by the following process. Detach from the column of 

 mercury a portion which will occupy about 40 or 50 degrees of the 

 scale : bring this column within the known part of the scale. Let 

 a, b be the readings at the upper and lower ends respectively ; a, ft 

 the index errors at these points as determined by comparison with 

 a standard. Move the column until its lower end coincides with 

 some degree below 32°, the upper end being within the compared 

 portion of the scale. Let c, d be the scale-readings for the upper 

 and lower ends in the new position, y being the scale error corre- 

 sponding to c. The error of the scale at d will then be 



d—[c — y—{a — a — b—ft))- 

 The true length of the detached column may be obtained with in- 

 creased accuracy by taking a mean of several measures within the 

 known part of the scale. This method was adopted for " Newman 

 (Makerstoun) " and " Troughton and Simms (Royal Society)," and 

 the following errors obtained : — 



Newman (Makerstoun). Troughton and Simms (R.S.). 



Temperature. Error. Temperature. Error. 



