34 Proceedings of the Royal Irish Academy. 
drical basin as a recipient ; the weight of this is, therefore, also reduced, 
as well as that of the connterpoise of the column. 
The length of the column capable of insuring satisfactory indica- 
tions can hardly be determined d priori. The greater this length, the 
larger the angle through which the column may move, and conse- 
quently the more extended the arcs to be described by the index point, 
as marking the amount of variation ; but, practically, there can only be 
a certain length of tube which will be safe for handling and working 
with ; while the drawing of very long tubes of a small diameter would 
perhaps present considerable difficulties. A tube of about T", 50, say 
5 feet, would probably be about the limit that might be usefully em- 
ployed. 
The limit of length at once fixes the limit of angle of oscillation of 
the column, since for this limiting position the mercury will completely 
fill the tube. 
This angle of oscillation at once fixes that of the index. 
This again must evidently count as part of the counterpoise, and 
consequently its weight is so far defined. Its length may, however, 
be evidently increased within limits corresponding to this weight. Its 
extremity may thus be made to describe an arc multiple of that de- 
scribed by the summit of the column. And, as this result can be 
attained without necessarily throwing additional weight on the bear- 
ings, it presents a decided advantage, as far as distinctness of observation 
is concerned. 
The angle of limiting position, as regards the vertical, is evi- 
dently one which must depend very much on t5e total weight act- 
ing on the bearings of the axis. Por positions very near the ver- 
tical (say 10°), and a low pressure of the atmosphere, the tendency of the 
column to incline under an increasing pressure would be comparatively 
slight ; or, in other words, the column would be sluggish. Por a greater 
angle (say 30°), the action of an increasing pressure would be more 
marked, the leverage being greater ; but the total amplitude of oscil- 
lation would be so far reduced, the length of the tube being taken as 
limited. 
The consideration of those several influences leads me to submit 
the following modifications for the construction of a barometer to 
work on the proposed principle. 
The tube to be conical, not cylindrical. Evidently any other varia- 
tion from the cylincbical form could with difficulty be practically 
realized, whilst a conical tube can be drawn to answer the required 
conditions of diameter. Por the portion which would correspond to 
the ordinary vertical height of the mercury, the cylindrical form 
might be retained, since the volume of mercury occupying this space 
remains constant for all positions ; the remainder of the tube to be 
made conical. 
The degree of tapering or conicalness should be such as would 
insure regularity of movement of the column, independently of the angle 
of inclination from the vertical. "With a cylindrical tube, the displace- 
