404A. WILMER DUFF, [VOL. IV.their own reflected images. If with better protection against air currents, etc., it is possible to work with a free disc D, the electrometer would remain absolute in its registry. The sensitiveness of such an apparatus would probably be of the order of o.ooi volt for a minimum of dielectric thickness.If the absolute character of the instrument be sacrificed in favor of greater sensitiveness, one may profit by the above experience, and replace the movable system DSE by a sheet of gold leaf made to cover the whole space on the inner face of the guard ring smoothly. The mirror E could then be reduced in size so as to weigh not much above a milligram. In such a system the counter force would be in part flexure and in part gravity. The former may be evaluated by the aid of the equation where x is the central displacement due to the pressure of/ dynes per square centimeter, r the radius of the disk (2.8 cm.), and s its thickness (say 0.0001 cm.). The pendulum forces are found as above. The results show that in this case x = F/4 nearly, where F is the total force on the disk, as given in § 2. Hence such an instrument should in the extreme case register as small a potential as 100 microvolts, supposing d= 0.01 cm., and the motion of one-tenth of an interference fringe discernible.Finally, the best conditions are secured by measuring the counterforces not by gravity, but in the usual way, by torsion. I have made an estimate based on the above experiments, supposing that two identical disks like D, in Fig. 1, are symmetrically joined by a light rigid rod, 20 cm. long, and suspended from a fine wire, 0.01 cm. in diameter, attached to the middle of the rod, with the ends fixed 10 cm. above and below it. In such a case, if ^=0.01 cm., the motion of one-tenth interference fringe would correspond to less than 20 microvolts. Whether this sensitiveness can be attained or not will depend very largely on quiet surroundings. Beyond this there does not seem to be more serious difficulty than was encountered in the above experiments carried on with very ordinary facilities. S EVERAL formulae for the viscosity of liquids at different temperatures have been proposed. The writer has recently required the use of wch a formula for a wide range of variation of viscosity. The following