320 PROCEEDINGS OF THE AMERICAN ACADEMY. 



such preliminary excursions were sufficient; after this the cylinder 

 settles down into a state in which the normal hysteresis cycles are 

 retraced with perfect regularity. Of course it is necessar}^ to make 

 the compressibility determinations immediately after this seasoning, 

 as the effect gradually disappears with time. The time occupied in 

 making the final readings to 12,000 kgm. and back with increasing 

 and decreasing pressure, making in all 20 readings, might vary from 

 two to three hours. After every change of pressure it was necessary 

 to wait for the temperature effect of compression to disappear; this 

 time was from 5 to 7 minutes. 



If the thermal dilatation under constant mean pressure is to be 

 determined, the seasoning consists simply in taking the cylinder once 

 through the temperature range contemplated as well as through the 

 pressure range. A word of description as to the general procedure 

 in determining the thermal dilatation at constant mean pressure will 

 not be out of place. The general plan is to change the temperature 

 while the piston is kept invariable in position, and therefore while 

 the volume is also approximately constant. The rise of temperature 

 produces a rise of pressure, so that after the rise of temperature it is 

 necessary to bring the pressure back to the former value by with- 

 drawing the piston if the change of temperature has been an increase, 

 or advancing it if the change of temperature has been a decrease. 

 The amount, by which the piston is withdrawn, as also the new final 

 pressure, is noted. The temperature is then changed again, and the 

 same set of readings made again. Thus every observation at any 

 given temperature involves two readings of the position of the piston 

 and the corresponding pressure. The slight change of pressure during 

 the changes of temperature carries with it hysteresis effects, which 

 it is necessary to avoid by previous seasoning, exactly as for pressure 

 changes over a wider range. Two processes of seasoning are necessary 

 for temperature, therefore, one a larger one for the entire temperature 

 range, and another smaller one for the slight changes of pressure 

 incident to the changes of temperature. This second seasoning is 

 made after the first more extensive seasoning simply by running the 

 pressure back and forth several times through the small range of 

 pressure to be met with during the temperature changes. This small 

 range was determined by preliminary experiment. 



In the actual calculation of the results there are a number of 

 corrections to be applied. These will now be discussed in detail 

 separately. In the first place the temperature coefficient of the 

 manganin coil has to be determined with particular care. This is 



