TRANSACTIONS OF SECTION A. 545 



chiefly to the fact that the amount of absoi-ption by the walls of the gauge 

 changes slowly when the volume of the vapour is changed by the compressiou 

 utilised in that inatvument. 



In both forms of our gauges, a mercurial siphon gauge, having tubes of five 

 centimetres diameter, is mounted on an instrument like a level-trier, and differ- 

 ences of level in tlie two sides are measured by determining the inclination of the 

 whole gauge which is required to bring the two surfaces to coincide with two 

 fiducial points in the axes of the two arms of the gauge. From the measured 

 inclination, together with the known linear distance of the two fiducial points, is 

 computed the diflference of level of the two surfaces of mercury. 



This principle (due to M.) has been carried out in two ways. In the first, the 

 siphon gauge is carried on a kind of bridge, supported at one end by two points 

 which rest on a horizontal plate on a solid pier ; and, at the other, by the point of 

 a micrometer screw, which itself rests on the same horizontal plate. In the axes 

 of the two limbs of this gauge are two platinum points, at the same level. The 

 amount of mercury in the gauge can be changed by a fine adjustment. 



When the pressures in the two arms of the gauge are the same, we determine 

 the zero reading. The amount of mercury in the gauge is altered till one fiducial 

 point barely touches the mercury, while the other creates a depression. Then the 

 inclination of the bi'idge is changed till the two depressions become equal. Mer- 

 cury is now removed from the gauge, when one depression will commonly disap- 

 pear before the other. The adjustment is repeated till both depressions disappear 

 together, or till both are apparently equal when made as small as can be seen. The 

 reading of the micrometer screw now is the zero reading, and marks when the two 

 points are in the same horizontal plane. 



If now the pressures in the two parts of the gauge become unequal, their 

 difference can be measured by determining what new inclination must be given 

 to the bridge and gauge in order to bring the two fiducial points into coincidence 

 with the merciuy surfaces again. Knowing the linear distance between the 

 fiducial points, we can compute their difference of level in their new position, and 

 so measure the difference of pressure between the two sides of the gauge. 



No optical appliances are needed in the use of this form of gauge. The ob- 

 server, moving his eye up and down, causes the image of a window bar to move 

 across the depression in the mercury made by the fiducial points. From the ap- 

 pearance of this image, he can, even without the aid of a magnifying glass, equalise 

 the depressions with a mean error less than the five-thousandth part of a milli- 

 metre ; after some practice, of course. But an observation requires two, three, or 

 four minutes. 



We have therefore constructed two gauges of a second form, employing the 

 same general principle, but also utilising an optical appliance (due to B.) by which 

 a reading is made as speedily as is an ordinary micrometric reading, while the 

 accuracy attained is even increased. Between the two arms of the siphon gauge 

 with its wide tubes is placed a pair of mirrors, so adjusted that the two fiducial 

 points, as well as the two images of these in the mercury, are seen side by side in 

 the field of a microscope carried on the apparatus and moving with the tubes and 

 mirrors. The surface of the mercury is not seen ; the two pairs of images of the 

 points, one belonging to the right arm of the gauge, and one to the left, are par- 

 tially superposed, so that the extreme ends of the points are perhaps a tenth of a 

 millimetre apart. If now the two real points are equidistant from the surfaces of 

 the mercury, the two pairs of images will seem equidistant ; if not, the inclination 

 of the whole system of gauge, mirrors, and microscope is changed till the distance 

 between the left-hand pair seems equal to that between the right-hand pair. This 

 is as easy as the bisection of a point with the wire of a micrometer. 



Mendel6ef found it necessary to grind and polish the external and internal sur- 

 faces of the glass tubes of his gauges, in order to eliminate errors due to irregular 

 refraction through irregular surfaces. In our apparatus the points of the tube 

 through which the fiducial points are viewed are always rigorously the same. We 

 therefore need only to secure an area in each tube through which we can get suffi- 

 ciently good definition ; it is easy to select such an area in the tube which is about 



