326 PROCEEDINGS OF THE AMERICAN ACADEMY. 



that the mercury was forced through solid steel by comparatively low 

 pressures at sufficiently high temperatures. 



In measurements undertaken by the author of various physical 

 constants at high pressures, this question of the action of mercury and 

 steel became of vital importance. For instance, the methods adopted 

 to measure compressibility assumed that there was no penetration of 

 the mercury into the steel containing vessel, as do also the methods 

 used more recently in determining the variation with pressure of the 

 freezing temperature of mercury and its change of volume on freezing. 

 The preliminary work at low pressures made it seem probable that at 

 least over the pressure range used by Amagat the effect described by 

 him does not really exist in the grades of steel used by him, and that 

 the observed eifect was due more likely to flaws in the steel. At the 

 same time it was found that at higher pressures there is undoubtedly 

 an effect important enough to demand the redesigning of the appa- 

 ratus for the measurement of the change of volume on freezing. It is 

 the purpose of this paper to describe the various experiments made to 

 prove the undoubted existence of the effect, and to offer a qualitative 

 explanation. The effect was run across only incidentally, and it was 

 examined only so much as was necessary for the work in hand. No 

 endeavor has been made to make the experimental investigfttion or the 

 explanation complete, as this would lead too far afield. 



The effect was first found during an attempt to measure the change of 

 volume of mercury on freezing by a method similar in many respects 

 to that of Tammann.^ It was found that cylinders of hardened 

 chrome nickel steel would support very much less internal pressure 

 when this pressure was transmitted by mercury than when the trans- 

 mitting fluid was some other liquid such as water. The pressure 

 might be less in the ratio of three or four to one ; thus cylinders 

 which stood without breaking 24000 atmospheres when the pressure was 

 transmitted to the interior by a mixture of water and glycerine broke 

 on the next appEcation of pressure at 5-8000 atmos. if the transmitting 

 fluid were mercury. These few preliminary experiments under vary- 

 ing conditions made the existence of an effect seem probable, but 

 pointed to nothing conclusively. It might well be that there was a 

 flaw running the entire length of the steel bar from which all these 

 pieces were cut, into which the mercury forced its way in consequence 

 of its greater mobility, in preference to the water, or it might be that 

 there was here a fatigue effect, the steel breaking more readily on the 

 second application of pressure with the mercury because of the ex- 

 ceedingly high pressure to which it had been previously exposed by 



' Tammann, Kristallisieren and Schmelzen (1903), p. 204. 



