1896.] Mr Griffiths, On Thermometry "Fixed Points!' 229 



experimental values as revised by Moritz. (The barometer read- 

 ings are corrected for temperature and for g to sea-level latitude 

 45°.) 



The mercury standard barometer used has been recently com- 

 pared with a Tonnelot barometer in the possession of Mr Heycock 

 which was standardised by Dr Guillaume in the spring of this 

 year, and I may remark that a discrepancy of as much as 2 mm. 

 would appear to exist between the Sevres and Kew standards if 

 the comparisons of my f-inch barometer are to be trusted. The 

 conditions of the comparison with Mr Heycock's barometer, how- 

 ever, were not entirely satisfactory, and I hope to have an 

 opportunity of repeating them under more favourable circum- 

 stances. I cannot therefore speak with certainty as to the absolute 

 value of the pressure which, however, is immaterial for our present 

 purpose. 



An error of "03 mm. in a barometer reading corresponds to 

 o, 001 in the boiling point, and great precautions, therefore, are 

 necessary to ensure sufficient accuracy. If the barometer is con- 

 tinually handled when resetting during a series of observations, 

 the thermometer attached to the instrument does not truly 

 indicate the temperature of the mercury in the barometer tube. 

 The compensated sulphuric acid barometer of Professor Callendar 

 and the writer's does not require handling, its scale is nearly six 

 times as open as the mercury scale, and its readings are inde- 

 pendent of temperature changes. It cannot, however, be relied on 

 for absolute readings, one reason being the permanent shrinking 

 of the glass bulb which is similar in nature to that characteristic 

 of mercury thermometers. The plan I have adopted, therefore, is 

 as follows. At regular intervals of about 20 minutes a careful 

 reading of both barometers is taken, the mercury barometer being 

 left untouched with the case shut throughout the interval, while 

 at intermediate times the sulphuric acid barometer is read through 

 a telescope and the absolute values deduced by interpellation from 

 the comparisons at each end of the interval. 



The following arrangement ensures great accuracy in the 

 setting of the ivory point, which is a frequent cause of error in 

 mercury barometer readings. The cistern is entirely surrounded 

 by a covering of black paper, with the exception of a small window 

 in front and another behind the ivory point. The point of view 

 is thus always the same. Across the further window is placed 

 a very thin black line whose image is seen in the mercury 

 across that of the ivory point. This image is viewed through a 

 powerful magnifying glass and the curvature of the line can be 

 seen before the dimple is apparent. I would strongly recommend 

 this method, for I find that great accuracy results from its 

 adoption. 



