486 jVL Hermann Herwig on the Independence of Temperature 



respect to the calculation, on the one hand as much too little 

 mercury in play as was pressed out by the larger lower half of 

 the bulb, and, on the other hand, that quantity of mercury too 

 much which upwards from the cross section was still present. 

 If we now try to equalize as nearly as possible these two quan- 

 tities in respect of their difference of temperature, we can entirely 

 eliminate the inaccuracy hence arising. For this purpose I have 

 for the average proportions of my experiments supposed the 

 above-mentioned cross section at about one third of the height 

 of the thermometer-bulb from above, and fitted and brought near 

 the firmly cemented iron end of the experimental tube (the bulb 

 was, besides, nowhere in contact with iron, but everywhere with 

 mercury ; it was, however, in its upper parts with suitable close- 

 ness surrounded by the iron) so that the equalization was as 

 nearly as possible attained. I have taken the greatest error 

 which could be made in this equalization, and with it calculated 

 that even in extreme cases in experiment the effect upon the 

 determination of the temperatures m (which are to be presently 

 discussed) would amount to less than a hundredth part of a 

 degree. Consequently the indications of the thermometer can 

 indeed be regarded as the upper temperatures of the mercury- 

 tube in the sense of the calculation. 



Now, before the commencement of the experiments this ther- 

 mometer was carefully and in detail compared in an oil-bath 

 with an air-thermometer, and, indeed, since it would subse- 

 quently always be made use of with a protruding thread, like- 

 wise in this position. At the close of the experiments I again 

 controlled the thermometer, and found it unaltered. I had 

 moreover, before using the thermometer, repeatedly and conti- 

 nuously heated it up to its highest temperature in order to give 

 it a definitive state. The temperatures which are subsequently 

 given according to this thermometer are, as has been said, directly 

 temperatures of the air-thermometer. As the thermometer was 

 graduated only to whole degrees, the reading could be accom- 

 plished to about o, 2. 



The other temperature to be determined on the experimental 

 tube itself was the mean temperature m. For its determination, 

 a mercury- tube could be arranged either as an outflow or as an 

 ordinary thermometer. At the commencement I preferred the 

 former. To this end a very fine outflow-aperture was made in 

 the upper iron closure of the tube, which permitted the mercury 

 to issue in very minute drops. It was further requisite to de- 

 termine the capacity of the tube at 0° and the coefficient of ex- 

 pansion of the glass. This latter was shown by a number of 

 measurements to be 0*000028 cubic per degree Celsius, there- 

 fore somewhat greater than usual. With these data and the 



