Pressure- Gauges for the Highest Vacua. 89 



Thus Crookes's data for air show the principles of both the 

 new gauges to be sound, but to get the best results from 

 each it ought to be developed on its own lines. To avoid 

 prolixity we will call the viscosity-meter when used as a 

 pressure-gauge the Viscometer Gauge ; its design on the 

 large scale has been worked out by Maxwell (Phil. Trans. 

 1866), who also furnished the theory of it, and on a smaller 

 scale by Kundt and Warburg (Pogg. Ann. clvi.). A still 

 smaller scale would, in most cases, be advisable for the 

 viscometer gauge, as its volume ought to be kept small. In 

 Maxwell's apparatus three glass plates 28 centim. in diameter 

 and 2 millim. thick were fixed at equal distances along a 

 solid axis, the continuation of which was a fine vertical 

 torsion- wire of steel ; fixed disks were mounted so that each 

 of the movable disks was half-way between two fixed ones 

 at distances varying from 5 millim. to 17 ; the whole was 

 enclosed in a glass vessel with a vertical continuation for the 

 125 centim. of torsion-wire. The inconvenience about this 

 form is that special arrangements have to be made for deter- 

 mining the viscosity of the rather strong wire that is required 

 to support so heavy a system. Kundt and Warburg sim- 

 plified the apparatus by reducing the weight and adopting a 

 bifilar suspension whose viscosity effect was practically negli- 

 gible. They used a single movable plate of glass of 16 centim. 

 diameter and thickness about 1 millim. vibrating between two 

 other fixed ones, the distance between fixed and moving 

 surfaces varying from PI millim. to 2*8. It is to be re- 

 membered that both these instruments were designed to 

 obtain accurate absolute values of the coefficients of viscosity 

 of gases, and for that purpose Maxwell worked out a pretty 

 complete theoretical expression connecting decrement of the 

 logarithm of consecutive angular amplitudes of the vibrating 

 system with the viscosity of the gas between the plates, the 

 viscosity of the wire, and the constants of the apparatus. 



Now we have just seen that Crookes succeeded in getting 

 remarkably good readings of log. dec. with his apparatus, 

 where the surface at which the viscosity produces its effect 

 is only the minute thickness of a small straight edge of a 

 mica plate moving not particularly close to a spherical sur- 

 face, that is to say, under very unfavourable circumstances ; 

 it is therefore evident that a thin glass, mica, or untarnishable 

 metal disk about 5 centim. in diameter and as light as pos- 

 sible, vibrating between two fixed plates with a distance not 

 greater than about 1 millim. between fixed and moving 

 surfaces, the suspension being by means of a quartz thread, 

 would give values of log. dec. of far higher accuracy than 



