752 TRUEBLOOD. 



shown in Figs. 5 and 6 were adopted only after a considerable amount 

 of experimenting, and after other forms (in particular, one in which 

 •a part of the tube between either plug connection and the correspond- 

 ing reservoir was vertical) had proven themselves unsatisfactory. 

 With these earlier forms, it was not found possible, even Avith the most 

 careful attention to filling the entire connections with water before 

 starting a run, to prevent the formation of troublesome bubbles of 

 air or steam in the parts of the tubes lying between the reservoirs and 

 the plug. The presence of these bubbles was indicated by a failure 

 of the mercury- columns to equalize in height on shutting the low side 

 flow valve VL; they were localized in the parts of the connections 

 mentioned by opening the test valves Vi, V^' in the cross-connection 

 ai 02 (Fig. 5) and closing one of the valves Fi, T'o, thus applying the 

 pressure existing on one side of the plug to both sides of the gauge 

 through the parts of the connections lying between the cross-connec- 

 tion oi 02 and the gauge. No differences of level of any significant 

 amount were ever observed when this was done. 



The upper cross-connection 6i 62 (Fig. 5) did not exist in the earlier 

 types of the connections between the plug and the manometer, but 

 was added when the type shown in Fig. 5 was adopted. It was added 

 because it was found that a part of the difference in level of the mer- 

 cury columns which persisted after closing the valve VL could not be 

 ascribed to the presence of air or water vapor in the upper connections, 

 but was due to a lag of the equalization of pressure on the two sides 

 of the plug behind the more or less regular pulses of the high side 

 pressure caused by the action of the piezostat: the throttling in the 

 plug, particularly in the case of radial flow plugs, pre\'ented immediate 

 equalization of the pressure following a pulse, and the upper cross- 

 connection, by short-circuiting the plug, removes the cause of this 

 lag. All these refinements in what would seem to be a comparatiAcly 

 insignificant part of the apparatus were necessary because the differ- 

 ence of pressure which persisted after stopping the flow of steam had 

 to be applied as a correction to the pressure difference observed dur- 

 ing a run. With the earlier forms of connections, this correction was 

 relatively large, being frequently in the neighborhood of 0.5 cm. of 

 mercury and occasionally as much as 1.0 cm. or even more. It was, 

 moreover, uncertain, being frequently different at the end of a run or a 

 series of runs from what it had been at the beginning. With the type 

 of connection shown in Fig. 5, the correction was reduced to so small 

 a figure that it could almost be ignored, for it now ranged from zero 

 to 1 or 2 mm. of mercury at the most (averaging about O.S mm.) — 



