\)\l. M. AV. TKAVKKS, ^] [;. (4. SEXTEK, AXD 1)\L A. .lAl^UEEOD 
1 -io 
'I'he iiiaiiiier in which this was ejected is sliown in tig. 4. Tlie dead-space and 
inanoineter culuinn were enclosed Ijetween t\^■o pieces of ])late-glass, ]'> and a, the 
j)ieces of [)late-glass p in front ffn ining the scale ; strips of wood, /• and r, formed 
the side and bottom of the water jacket, the whole being made water-tight bv means 
of a mixture of red-lead and o'old size. The arrano’ement was carried on brass 
brackets ss screwed to the face of the main snp])ort. The two plates of glass M’ere 
clamped to the wooden strips rr' Ijy brass bands ?dand wooden wedges mu'; the brass 
bands were screwed at one etid to the main siipp)ort. Water entered the jacket at the 
bottom, and esca]nng at the top by a rubber tube, flowed over the tubes li and k, and 
the ])ortion of (j below tlie water jacket. This })art of the ap})aratiis was swathed 
in cotton cloth, and was maintained at a constant, though indefinite temperature, 
by means of the cuiTent of water. Tlie temperature of the water in the jacket 
remained constant within 0‘^'02 for a sutficient time for the measurement of the 
pressure, and rarely varied by 0°'l during the period of one hour, necessary for four 
consecutive observations. The error due to tem])erature on Tjqq (950 millims.) is 
thus i-educed to 0‘002 millim. vhich is considerably less than the error of observa¬ 
tion. It may also l)e noted here that, as the temperature of the dead-space was 
made wu’th the same degree of accuracy, and an error of OT in determining it makes 
a difference of 0 002 millim, on P,,,,,, this source of error also disappears. 
The height of the column of wliich the tem])erature varied l)etween 8° and 17'^ (h 
was reduced in everv case to the heiold at fO'' ( 
It is obvious that as the manometer tube and the glass scale in front of it are not 
])arallel, and tlie space between them is filled with water, a refraction error may be 
introduced which will have a constant influence on the value of Pq and Pjoo- 
de^’iation of the manometer tube amounts to 2 millims. in its total length, the error 
in reading the height of the column would he roughly O'Ol millim. By oh.serving 
the height of the column wfiien the jacket contained waiter, and then allowung the 
water to escajie rajiidly, w'e could, how'ever, detect no change in the position of the 
meniscus, Furtlier, since the manometer tube was sufficiently straight for the error 
t(^ atfect P|, and P,,,,) eipially. the error due to i-efraction can he neglected. 
15. Th<‘ McasurcriK-ul af tin' l^ressiire on the tjos in the Thennometer 
The observations \\ere made by means of a telesco])e placed at a distance of one 
metre from the scale. The telescope wms fitted with a llilger ocular micrometer, 
w 1th a screwp divided into 100 parts, each corresponding to a movement of the cross¬ 
wire of ()'005 minim., or to 0'007 millim. on the scale. Each meniscus was 
illuminated from hehind hy means of an electric glow -lamp covered with white tissue 
jiajier; cards ])laced behind the glass jilate O, cut off the light at a height of 1 or 
2 millims. above the level of the mercui’v {of. 1>. 109). 
