and narrow cylinder will be used.  Such a system, however, can only be employed when the cylinder is surrounded by spring water, since, without that, an intimate contact with earth that is only moist, cannot be obtained with certainty.  In earth that is only moist and for moderate depths, preference may be given to an electrode laid down flat.  The digging necessary in this case is onerous, it is true, but it permits of very accurately determining the state of the earth beneath and of obtaining a very perfect adherence of the electrode therewith.  Two forms, the annular ribbon or the flat ring and the network, present themselves, according to calculations, as a substitute for copper plates, which are so expensive; and these forms are satisfactory on condition that the labor of digging be not notably increased.  These forms should always have a diameter a little greater than that of the plate.  The flat ring and the network, however, offer one weak point, which they possess in common with the plate, and that is, their dimensions cannot be easily adapted to the nature of the ground met with without a notable increase in the expense.  Now, if the ground should offer a conductivity less than what was anticipated, and it were desired to increase the plate, say by one-third, it would be impossible to do so as a consequence of the closed form.

One important advantage is realized in this respect by combining the ring and the network in the form of a reticulated ring having a diameter of from 1 to 11/2 meters.  On cutting this ring at a given place and according to a certain radius we obtain the reticulated ribbon shown in the accompanying figure.  The thickness of the wires is 2.5 mm., and their weight is 0.475 kilo. per meter.  L, L, and L are the points at which the conducting cable is soldered.  A reticulated ribbon of copper can be made in advance of any length whatever, and, according to local exigencies, it may be easily curved and given the form of a flat or cylindrical ring of varying width.  Even though the ribbon has already been cut for a ring of given diameter, it may be still further enlarged by drawing it out and leaving a bit of the ring open, so as to thus obtain a nearly corresponding diminution in the resistance.  Such a resistance may be still further diminished by rendering the ring higher, that is to say, by employing an annular cylindrical form.

After assuring himself, by experiments on a small scale, that calculation and observation gave concordant results for the flat ring, the author made an experiment on a larger scale with the annular network.  For practical reasons he employed for this purpose a copper wire 2.5 mm. in diameter, which may be expected to last as long as one of iron plate 2 mm. in thickness.  Calculation showed that in a ribbon 160 mm. wide, meshes 40 mm. in breadth were advantageous and favorable as regards rigidity.  A reticulated ribbon like this, 4 meters in length, was made and formed into a flat ring having an external diameter of 1.42 m. and an internal one of 1.10 m.  The resistance of this ring was found to be W = 0.3485 (1/_k_), and that of a plate one meter square, W0 = 0.368 (1/_k_).