258 



NA rURE 



[July 15, 1897 



in all its bearings. Needless to say, such a committee should 

 include representatives of both the " splitting " and " lumping" 

 interests ; and if the points at issue wefe fairly debated, with 

 a full determination to give and take on both sides, it is diffi- 

 cult to believe that a working compromise between the extreme 

 views could not be arranged. Almost anything is better than 

 the present condition of uncertainty and discrepancy. 



R. Lydekkkr. 



RECENT INVESTIGATIONS INTO THE 



NUMERICAL VALUE OF 



" THE MECHANICAL EQUIVALENT:' 



'T'HE value of the "mechanical equivalent," when deduced 



from experiments based on the direct transformation of 



mechanical work into heat, affords the best standard by means 



of which to test the validity of our system of electrical units. 



It is evident, however, that the value of this test depends 



upon the accuracy of the "equivalent" determinations. The 



engineer may (very rightly) regard extreme numerical accuracy 



in this case as comparatively unnecessary, but from the 



physicist's point of view there are few natural constants whose 



exact determination is of equal importance.^ 



Until the present time the evidence available has been so 

 conflicting that it has been impossible to draw any certain con- 

 clusions from a comparison of the beat 

 developed by mechanical work with that 

 resulting from work electrically performed. 



In Phil. Trans. Roy. Soc, 1893, I gave 

 an account of an investigation, by electrical 

 methods, into the capacity for heat of 

 •water. The chief object of that work 

 was to apply to the electrical units the 

 test above referred to, for I considered 

 that Rowland's admirable series of ex- 

 periments {Proc. American Academy, 1879) 

 on the direct conversion of mechanical 

 energy supplied sufficient data to render 

 such an investigation desirable. I regret, 

 however, to say that, for the following 

 reason, the results of that work have 

 hitherto been of little value for the par- 

 ticular purpose for which it was under- 

 taken. The change in the capacity for 

 heat of water indicated by Rowland differed 

 materially from that obtained by means 

 of the electrical experiments. This differ- 

 ence in the rate of change must be due to 

 differences in temperature measurements. 

 In Phil. Trans., 1S93, p. 496, I wrote as 

 follows: "No change in the value of the 

 various units or constants involved in our in- 

 vestigations could bring our results into absolute agreement with 

 those obtained by Rowland, since, owing to the difference in the 

 expressions for the temperature coefficients for the specific heat 

 of water, it is inevitable that if our conclusions should agree at 

 some one temperature, they must necessarily differ when ex- 

 pressed in terms of a thermal unit at any other temperature, 

 and thus changes in the values of the units would only alter the 

 temperature of agreement." 



This quotation will, I think, render evident that but little 

 progress could be made until some explanation of the dis- 

 crepancies in the temperature measurements was forthcoming. 



An indirect comparison of Rowland's standard with that of 

 the Bureau International is given in Prof. Schuster's paper on 

 the "Scale value of Dr. Joule's thermometers" {Phil. Mag., 

 1895). The results indicate that Rowland's rate of decrease 

 in the heat capacity of water would be diminished if expressed 

 in terms of the International Standard ; but, as Prof. Schuster 

 remarked, " it would be necessary to have further information 

 before any definite conclusions could be drawn." 



I am glad to say that we are now in possession of the further 

 information sought for by Prof Schuster, and the above brief 

 statement of our difficulties has been made in the hope of 

 drawing attention to the new light now thrown on the whole 

 subject. 



1 It has been decided (see Report of the Electrical Standards Committee, 

 1896) that the thermal unit is to be a dynamical one, hence the demand for 

 accuracy becomes insistent. 



Two entirely distinct investigations have just been broughi 

 to a successful conclusion in the laboratory of the John> 

 Hopkins University. 



(i) " A recalculation of Rowland's value of the mechanical 

 equivalent of heat in terms of the Paris hydrogen thermometer, ' 

 by W. S. Day. 



(2) "A comparison of Rowland's mercury thermometer witli 

 Griffiths' platinimi thermometer," by C. W. Waidner and F. 

 Mallory. 



Both the above investigations have been carried out under 

 the directions of Profs. Rowland and Ames. Full particulars 

 of the work will shortly be published in America ; but, in the 

 meantime, the authors have very kindly given me permission 

 to publish the results in this country. I will here give no 

 details beyond the statement that the comparison of Rowland's 

 thermometers with those of the Bureau International were made 

 under conditions as nearly as possible similar to those prevalent 

 during Rowland's experiments, and the same remark holds 

 good with regard to the comparison with the platinum standard. 



The results of these entirely separate investigations may be 

 briefly summed up as follows : — 



(i) The values resulting from Rowland's experiments undergo 

 considerable modification at certain temperatures. 



(2) Over the temperature range covered by Griffiths' experi- 

 ments (14° to 26° C.), the rate of change in the capacity for 



TEMPERATURE ON THE PARIS HYDROGEN SCALE. 



heat of water becomes practically identical with that given by 

 Griffiths. 



(3) Throughout this range, Griffiths' value exceeds Rowland's 

 by about i in 420. 



(4) Separate standardisations of the same platinum thermo- 

 meter were performed both in England and in America. The 

 units adopted in the two cases differed slightly, but this is un- 

 important, as the temperature measurements are independent of 

 the magnitude of the unit. The essential point is the respective 

 values of the ratio of the resistance at 100" C. to that at 0°. 

 These were as follows : 



English 

 American 



= 1-38596 

 = I -38597 



Thus affording satisfactory proof that not only the electrical 

 measurements, but also the barometric standards, &c., are in 

 perfect agreement. 



(5) The results of the comparison with the platinum standard 

 are (in the words of the authors) "in almost absolute agree- 

 ment " with those deduced by Mr. Day from the direct com- 

 parison with the international standards, and thus the validity 

 of Callendar and Griffiths' method of standardising the platinum 

 thermometer is confirmed. 



In the reduction of Rowland's results, "each individual 

 experiment, the thermometers used in it, and the number of 

 observations made with each thermometer, were taken into 

 account.' 



NO. 1446. VOL. 56] 



