The separating effect of these partitions for the gas is wholly due to capillary phenomena.  We know, in fact, that water tends to expel gas from a narrow tube with a pressure inversely proportional to the tube’s radius.  In order to traverse the tube, the gaseous mass will have to exert a counter-pressure greater than this capillary pressure.  As long as the pressure of one part and another of the wet wall differs to a degree less than the capillary pressure of the largest channel, the gases disengaged in the two parts of the trough will remain entirely separate.  In order that the mixing may not take place through the partition above the level of the liquid (dry partition), the latter will have to be impenetrable in every part that emerges.  The study of the partitions should be directed to their separating effect on the gases, and to their electric resistance.  In order to study the first of these properties, the porous partition, fixed by a hermetical joint to a glass tube, is immersed in the water (Fig. 2).  An increasing pressure is exerted from the interior until the passage of bubbles is observed.  The pressure read at this moment on the manometer indicates (transformed above the electrolytic solution) the changes of level that the bath may undergo.  The different porcelains and earths behave, from this point of view, in a very unequal manner.  For example, an earthen vessel from the Pillivayt establishment supports some decimeters of water, while the porcelain of Boulanger, at Choisy-le-Roi, allows of the passage of the gas only at pressures greater than one atmosphere, which is much more than is necessary.  Wire gauze, canvas, and asbestos cloth resist a few centimeters of water.  It might be feared, however, that the gases, violently projected against these partitions, would not pass, owing to the velocity acquired.  Upon this point experiment is very reassuring.  After filling with water a canvas bag fixed to the extremity of a rubber tube, it is possible to produce in the interior a tumultuous disengagement of gas without any bubbles passing through.

[Illustration:  Fig. 2.—­Arrangement for the study of capillary reaction in porous vessels.]

From an electrical point of view, partitions are of very unequal quality.  Various partitions having been placed between electrodes spaced three centimeters apart, currents were obtained which indicated that, with the best of porcelains, the rendering of the apparatus is diminished by one-half.  Asbestos cloth introduces but an insignificant resistance.