Manchester Memoirs, Vol. xlii. ( 1 898), No. 0. 47 



{U) The index en-ors of the thermometers. 



For the initial temperature thermometer this error 

 was determined by immersing it in a mixture of crushed 

 ice and water. In most of the trials this thermometer 

 gave readings between 33° and 34°. On this interval the 

 index correction was found to be correct when compared 

 with a standard thermometer in the Physical Laboratory 

 of the College. Higher up the scale a slightly larger 

 negative correction to the readings was found to be 

 necessary, and this correction was applied to the later 

 trials when required. 



The main accepted experiments lasted from February 

 to July, 1896, and during that interval six determinations 

 were made of the index error of this instrument. 



The correction required by the readings fell from 

 — 0-48"' on Feb. 5th to —0-52° on July 7th. 



By the use of Regnault's steam tables, corrections were 

 obtained to the indications of the thermometer used in 

 the discharge pipe at the points 212°, 213-8°. and 215-6°, 

 and also at lower temperatures when the barometric 

 pressure permitted. 



The corrections were obtained by immersing the ther- 

 mometer in steam at these temperatures, the pressures 

 being reduced directly from Regnault's table. 



These pressures were measured to joVoiii- of mercury 

 by means of a combined manometer and barometer 

 specially constructed for the purpose. 



The corrections obtained fell from +0/24 on February 

 8th to +0-04 on July 6th, 1896. 



Both of these thermometers worked under a con- 

 siderable pressure, and on that account a determination 

 of the elevation in their readings, due to a known pressure, 

 was made. This elevation I found, in the case of the 

 thermometer (Q2) used in the supply pipe, to be 0-0072'^ 

 per inch of mercury pressure, and for the thermometer 



