32 Starkweather — RegnauWs Calorie and Our 



ing temperature, which is correct ; but the increase is so rapid 

 as to look a little suspicious. According to Battelli the increase 

 is very slow. 



Even if Ramsay and Young's experiments are substantially 

 correct far from the saturation line, they cannot be used, for 

 they contain a constant error, the incorrect determination of 

 the weight of water present in the tube. There is no agree- 

 ment with Battelli by which this error can be estimated, nor 

 can we determine it from our saturation volumes already 

 obtained, since in Ramsay and Young's experiments near the 

 saturation line great condensation occurred. 



We have left therefore only Battelli. In a table below is 



given a comparison of — according to such of his experiments 



as lie near the saturation line with that according to Regnault's 

 experiments, the volumes being calculated from the latent heats. 

 Each of the latter is given immediately under that one of 

 Battelli's which is to be compared with it, and the fractional 



deviation of — 



is given last. 











p 129-14 

 T 330-7 



330-78 

 352-2 



749-12 

 37333 



2060-1 

 404-01 



3061-9 

 417-94 



7971-4 

 456-6 



12181* 



475-71 



V Z 3-413 

 T 



3-411 



3-392 



3-373 



3-350 



3-276 



3-210 



p 148-79 

 T 333-7 



354-62 

 3537 



760 

 3737 



2030-3 

 403-7 



2717-6 

 413-7 



7546-4 

 453-7 



11689 

 473-7 



£ 3-420 

 T 



3-403 



3-373 



3-317 



3294 



3-196 



3-143 



d --002 



+ •002 



+ •006 



+ •017 



+ •017 



+ •024 



+ •021 



It is at once seen that the discrepancy is very bad. Were 

 the deviations negative instead of positive they could be 

 ascribed to condensation in the case of Battelli, and his experi- 

 ments not near the saturation line would still be available. Let 

 us now see, supposing Battelli to be correct, where an error in 

 our saturation volumes can be placed. We have seen that the 

 greatest error which can possibly be ascribed to the mechanical 

 equivalent of Regnault's calorie is 0*5 per cent. This would 

 make the deviations respectively 



— •007 — -003 +-001 +-012 +'012 +'019 +'016 



Further we have seen that the latent heat at 200° C. may be 

 at most 0*6 per cent too small, although at 100° O. it is correct. 

 Allowing errors in the latent heats the deviations become 



— •007 — -003 +*001 +-010 +-010 +-014 +'010 

 The two negative ones, at low pressures, could be ascribed to 



