OP ARTS AND SCIENCES. 179 



By this the correction for the radiation and the error due to the 

 irregularity of the engine are changed, and yet scarcely an appreciable 

 difference in the results can be observed. 



The tables explain themselves very well, but some remarks may be 

 in order. Tables XXXVII. to L. inclusive are the results of fourteen 

 experiments selected from the total of about thirty, the others not 

 having been worked up yet, though I propose to do so at my leisure. 



Table LI. pives the collected results. At the top of each coin inn 

 the date of the experiment and number of the thermometer are 

 given, together with the approximate torsion weight and the rate of 

 rise of temperature per hour. The dash in each column gives 

 approximately the temperature of the jacket, and hence of the air. 

 There are four columns of mean values, but the last, produced from 

 the combination of the table by parts, is the best. 



Table LII. gives the mechanical equivalent of heat as deduced from 

 intervals of 10° on Table LI. The selection of intervals of 10° tends 

 to screen the variation of the specific heat of water from view, but a 

 smaller interval gives too many local irregularities. In taking the 

 mean I have given all the observations equal weight, but as the Kew 

 standard was only graduated to J° F. it was impossible to calibrate it 

 so accurately as to avoid irregularities of 0°.02 C. which would affect 

 the quantities 1 in 500. Hence, in drawing a curve through the 

 results, as given in the last column, I have almost neglected the Kew, 

 and have otherwise sought to draw a regular curve without points of 

 inflection. The figures in the last column I consider the best. 



Table LIII. takes the mean values as found in Tables LI. and LII., 

 and exhibits them with respect to the temperatures on the different 

 thermometers, to the different parts of the earth, and also gives the 

 reduction to the absolute scale. I am inclined to favor the absolute 

 scale, using m = .00015, as given in the Appendix to Thermometry, 

 rather than .00018, as used throughout the paper. 



Table LIV. gives what I consider the final result of the experiment. 

 It is based on the result m^= .00015 for the thermometers, and is 

 corrected for the irregularity of the engine by adding 1 in 4000. 



The minor irregularities are also corrected so that the results signify 

 a smooth curve, without irregularity or points of contrary flexure. 

 But the curve for the work does not differ more than three kilo- 

 gramme-meters from the actual experiment at any point, and generally 

 coincides with it to about one kilogramme-meter. These differences 

 signify 0°.007 C. and 0°.0O2 C, respectively. The mechanical equiv- 

 alent is for single degrees rather than for ten degrees, as in the 

 other tables. 



