246 PROCEEDINGS OF THE AMERICAN ACADEMY. 



fact that his results agree with those of Peake and of Griessmann 

 shows that none were necessary in his apparatus. 



This agreement is in many other ways a significant one, for it is 

 inconceivable in view of the great differences in almost every respect 

 between the details of the three sets of apparatus, that any serious 

 systematic errors should have been present in any one of the sets of 

 results without completely destroying the agreement between them. 

 This is particularly true in the matter of heat insulation, where the 

 precautions taken by the three observers had almost nothing in com- 

 mon except effectiveness. In Dodge's work also this point was care- 

 fully considered but the results are not so satisfactory. They will be 

 discussed and a correction computed on page 262. 



In all four cases the thermometry is the weakest part of the work. 

 It is especially unfortunate for the present purpose that the original 

 aim of the experiments did not require or suggest that the difference 

 between the temperatures before and after the expansion be measured 

 as such, as by a thermocouple or a differential resistance thermometer. 

 The subtraction which must now be made of one reading on a mercury 

 thermometer from another reading on another thermometer, to give 

 a small difference, is not a particularly accurate method of getting 

 that difference. The same is true of the determination of the pressure 

 drop. The individual measurements were comparatively good, being 

 made in three of the cases with carefully calibrated Bourdon or spring 

 gauges, and in the fourth case by an extra measurement of the temper- 

 ature of resaturation of the low side steam, but the differences needed 

 in this paper must inevitably be subject to comparatively large errors. 

 The reader must therefore be prepared for much lack of self-consistency 

 in the results. It is hoped that the errors are largely incidental errors 

 such as can be eliminated by averaging. 



Grindley's experiments were performed in England during the winter 

 of 1897-8. His data are given in full in his paper and are plotted in 

 his Diagram 5 reproduced here as Figure 2. It will be observed that in 

 every case his steam drops several pounds in pressure before it leaves 

 the saturation line. This he explained by means of a curious and now 

 discredited "heat of gasification." A better explanation is that his 

 steam was initially slightly wet. Since this source of error affects 

 the high side data of every one of his experiments, it might seem that 

 all of his work must be rejected. It will be noticed, however, that his 

 experiments are grouped into runs ; that is, if in a certain experiment 

 steam in a certain initial condition has been throttled to a certain low 

 side pressure and temperature, then in later experiments of the same 

 group, steam in the same initial condition is more and more throttled 



