962 Mr. Beer and Prof. Tyndall on Manometric 



a profound modification of the convection current must Lave 

 been brought about by this means. Again, the carbons were 

 enclosed in a horizontal quartz tube 1\ in. in diameter and 

 closed at one end. Readings under these conditions were 

 very difficult to obtain, and it was almost impossible to 

 maintain a silent central burning arc. But, so far as they 

 went, they gave no evidence of modification as a result of 

 the prevention of convection by the tube. 



These results are consistent with the view that the pres- 

 sures have their origin in the arc itself, and arise from ionic 

 streams giving rise to a residual wind-pressure. Evidence 

 as to the particular region or regions in the arc in which 

 this wind is generated can be deduced from a study of the 

 variation of pressure with length of arc at constant current. 

 For this purpose the following observations were made. 

 It is convenient to deal with the anode and cathode results 

 separately. 



A node pressure. 



Values of pressure and length of arc for various currents 

 with holes of different sizes were obtained. For reasons 

 which will be clear later, only those for a hole of diameter 

 1*5 mm. are inserted in this account, and these are shown 

 graphically in curves III. 



It will be seen that fluctuations of considerable magnitude 

 occur ; these could be traced to slight departures from true 

 centreing on the hole of the luminous patch on the electrode. 

 In fact, the readings are so sensitive to this that much work 

 was necessary before we were assured that results such as 

 are given there represent the facts. 



With each observation great care was taken that the 

 luminous patch was centred as perfectly as possible. This 

 was easy with short arcs, but with long ones burnt in the 

 ordinary way there was a marked lack of centrality — due 

 no doubt to the action of convection currents on the flame, 

 as the arc was horizontal ; but by holding a quartz plate 

 horizontally just over the arc, it was found that the patch 

 could be brought back on to the hole, and "central" 

 readings for all lengths of arc were thus obtained. 



Curves III. show that the pressure at the anode is practi- 

 cally, if not wholly, independent of the length of the arc, 

 the apparent slope of two of the lines being not more than 

 might be attributed to experimental errors. 



This result strongly suggests that the pressure originates 

 in the region quite close to the anole itself, and experiments 



