483 



points. Af'ler tlie com|tletioii of the investigation described in the 

 preceding communication the various points could be each separately 

 checked by a comparison with the isothermal deduced from the 

 points combined. If the manometer had been fdled with a different 

 gas or an arbitrary mixture of gases, it would have served, its 

 purpose as an intermediaiy between pressure-balance and open 

 manometer equally well. 



Tal)le II contains the results of the comparison of the pressure- 

 balance with 7l/i2„, extending over the range from 60 to 1Ü0 atmos- 

 pheres. The measurements of March 27 and 29 and June 24 were 

 made with the air-liquid transmission of pressure, as had been those 

 with il/„, whereas in those of April 23 and 24 and June J 8 and 

 19 use was made of the liquid system mercury -glycerine-oil which 

 was arranged later on as described in one of the preceding sections. 



6. Discns.tion. The results of all the measurements as contained 

 in the above tables lead to the following conclusions: 



1. The functional section is not independent of the pressure, but 

 as the pressure rises above 20 atmospheres it increases, goes through 

 a greatest value at about 70 atmospheres and then diminishes with 

 greater rapidity as far as the comparison reached. The greatest 

 deviation is 0.0020. 



2. When the determinations were repeated, the same value was 

 not always found for the functioiuxl section, the greatest deviation 

 being about 0.0005 in this case. 



3. The functional section differs from the geometrical section as 

 given by Schafkkr and Budenberg (1 cm^) by about 0.0030. 



4. The sensitivity of the pressure-balance nrooo t''"S far exceeds 

 its accuracy. If the latter is to be raised to the value of the sen- 

 sibilit}', the theory of the instrument will have to be developed and 

 means will have to be found to obtain constant results within the 

 limits of the sensibility. Probably in order to attain this accuracy 

 a pressure-balance will always directly or indirectly have to be 

 compared with an open manometei'. 



5. Pressures which have been measured with a Schaffer and 

 Budenberg pressure-balance which has not been calibrated cannot at 

 present be estimated at a higher accuracy than about xoir, provided 

 that the error in the area of the piston is not larger tlian 0,1 "/„. 



In conclusion we wish to thank Professor Kamermngh Onnes and 

 Professor Kohnstamm for their sustained interest in our work. 



