trigonometrical measurements at Spitzbergen, 303 
Having brought down Newman’s barometer from the hill 
to the Observatory on the afternoon of the 21st, its direct 
comparison with Jones’s was resumed, for the purpose of as- 
certaining that it had sustained no injury, and to make an 
additional trial of the index correction and neutral point, as 
the height of the mercury was then much lower than in the 
former comparisons. The barometer had been progressively 
falling from the afternoon of the 19th, and continued to de- 
scend until between 5 A. M. and 7 h 30™ A. M. on the 22nd, 
when it reached 29.244 its lowest depression: it had fallen 
considerably, therefore, and was still falling when I began 
to compare the barometers on the afternoon of the 21st: the 
surface of the mercury in the tube of Newman’s, the dia- 
meter of which is only .15 parts of an inch, was convex as 
usual, though not so much so as when rising or stationary ; 
but in Jones’s, the diameter of which is double that of the 
other, or .31, the convexity had entirely disappeared, so that 
the zero of the vernier, when marking the level of the highest 
part of the mercury, coincided with the part which was in 
contact with the glass, and I was even doubtful whether there 
was not a slight concavity in the centre. I had not leisure at 
the time to examine the comparisons, but continued to repeat 
them at intervals, until the following morning, when, as the 
mercury began to rise, I became more strongly impressed 
than on the preceding evening, that it had been previously 
slightly depressed in the centre below the level of the sides, 
and that the depression had ceased to obtain. On apply- 
ing the respective corrections to the comparative readings 
contained in the subjoined table, it appeared that as soon as 
the mercury had began to rise, the index correction of -{-.196 
