March 30, 1899] 



NA TURE 



521 



100° C, is given in the followinj:; table, in which the rate of 

 increase is expressed I)y finding the power n of the absolute 

 temperature T to which the viscosity is most nearly proportional. 

 The most concordant results were obtained by the method of 

 transpiration, and gave an average of 76 for the index n in the 

 case of air. The more condensible gases gave larger values for 

 the rate of increase, but the value for hydrogen appeared to be 

 smaller. 



T.AELE III. — Variation of Viscosity v with Temperature T. 



Formula, vlv„ = (T/r„)". 



Observers, Dates. Values of Inde.\ « (o' to 100 C.) 



It will be observed that the results are not very concordant, 

 but the experiments are much more difficult and liable to error 

 than might be supposed. The most accurate method was that 

 employed by Holman, but even in this case the margin of un- 

 certainty is considerable. It would evidently be impos.sible to 

 employ the method of transpiration to any advantage for the 

 determination of temperature unless a far higher order of ac- 

 curacy could be easily attained. After repeating the majority 

 of the more promising methods in detail, including the original 

 method of Maxwell, the writer came to the conclusion that they 

 were entirely unsuitable for the purposes of thermometry, and 

 would have abandoned the attempt entirely if he had not for- 

 tunately succeeded in finding a more perfect way. 



Application of Electrical Analogies. 

 In studying the flow of electricity through conductors, which 

 is in many respects analogous to that of a fluid through a fine 

 tube, electricians have been compelled, from the intangible 

 nature of the fluid with which they work, to elaborate the most 

 delicate and powerful methods of investigation. One of the 

 most useful of these methods is generally known as the Wheat- 

 stone-bridge method, and is used for measuring the resistance 

 of a conductor to the passage of an electric current. The 

 method is equally applicable and equally exact for determining 

 the resistance of a fine tube to the passage of a gas. The writer 

 was already very familiar with the application of this method in 

 all its refinement of detail to electrical resistance thermometry. 

 The suggestion for applying it to the closely analogous problem 

 of transpiration was supplied by the researches of W. N. Shaw, 

 F.R.S., who had already applied it, in connection with certain 

 experiments on ventilation, to the effusion of air through large 

 orifices at ordinary temperatures. 



Shaw's Effusion Balance. 



The apparatus used by Shaw (described in the Proc. A'oy. Soc, 

 vol. xlvii., 1890) consisted of boxes to represent rooms, with 

 apertures about half a square inch in area to repre.sent ven- 

 tilators. Two of these apertures were made in the form of 

 adjustable slits. The circulation of air through two rooms in 

 parallel was maintained by a gas burner, and the slits were 

 adjusted to make the pressure in the two rooms the same, as 

 indicated by the absence of flow in a connecting tube, contain- 

 ing a pivoted needle and vane as a current detector. The balance 

 was shown to be independent of the air-current when that was 

 varied from one to four cubic feet per minute. The effusion re- 

 sistance of an aperture was also verified to be approximately 

 proportional to the square of the reciprocal of the area, with 

 apertures of similar shape. This method of investigation was 

 admirably adapted to problems in ventilation, in which the 

 phenomena depend mainly on effusion through relatively large 

 apertures. It would, however, be difiicult to adapt to the 

 problem of temperature measurement. It would not be easy to 

 make an aperture which could be continuously varied without 

 changing its shape, and at the same time to measure the change 

 of area with sufficient accuracy, if the area were small enough to 

 prevent appreciable cooling of the thermometer by the current of 

 air flowing through it. There is also the disadvantage that the 

 pressure-difference varies as the square of the current ; so that, 



NO. 1535, VOL. 59] 



if very small currents are used, the effects of viscosity become 

 more important, and the balance ceases to be independent of 

 the current, unless everything is symmetrical and at the same 

 temperature in corresponding parts. 



For these reasons it seemed preferable, in applying the 

 Wheatstone-bridge method to air-currents, to employ fine tubes 

 as resistances, and to eliminate the effects of effusion as com- 

 pletely as possible, at least in the resistance-measuring part of 

 the apparatus. With transpiration resistances the current is 

 directly proportional to the pressure difference, the electrical 

 analogy is much closer, and the theoretical conditions can be 

 very accurately realised. 



The Transpiration Balance. 



The Wheatstone-bridge method of measurement proved to be 

 so exact and .so perfectly adapted to the problem of transpir- 

 ation thermometry, that, after some preliminary experiments, 

 the writer had a very elaborate apparatus constructed, in the year 

 1893, which was in every detail the exact analogue of an elec- 

 trical resistance thermometer. The fine wire resistances of the 

 electrical apparatus, in terms of which the change of resistance 

 of the thermometer is measured, are replaced in the transpir- 

 ation box by a graduated series of fine tubes, which can be 



Fig. 4. — Diagram of transpiration balance. 



short-circuited by means of taps of relatively large bore, cor- 

 responding to the plugs of negligible resistance in the electrical 

 resistance box. The galvanometer is replaced by a rheoscope, 

 constructed after a pattern devised by Joule for a different pur- 

 pose, which can be made to rival in delicacy the best modern 

 electrical instruments. The pyrometer itself consists of a fine 

 tube of platinum instead of a wire, and is fitted with " compen- 

 sating leads" to correspond with those of the electrical instru- 

 ment. All the details of the methods of observation and 

 calibration are faithfully copied from the electrical apparatus, 

 and the result, so far as the measurement of transpiration 

 resistance is concerned, are equally satisfactory. 



Fig. 4 is a diagram of a working model of the transpiration 

 balance, which was exhibited at the lecture. This model has a 

 vertical needle for index, and a pivoted mica vane, which is 

 deflected when a current flows through the bridge piece. It is 

 constructed to work on the ordinary lighting-gas pressure, and 

 to give its maximum deflection for a 10 per cent, change of 

 resistance with the gas about half off". With all the taps off, 

 the resistances on either side are equal, and there is no deflection. 

 In the diagram the balance is supposed to have been disturbed 

 by opening one of the taps. The apparatus actually used for 

 temperature measurement has sixteen taps, and a mirror, 

 rheoscope, and is a thousand times more sensitive. 



