[LM111230], Letter from Arthur Edwin Kennelly to Edward Hibberd Johnson, December 1888


View document with UniversalViewer   → View document on Archive.org  → Re-use this digital object via a IIIF manifest


[LM111230], Letter from Arthur Edwin Kennelly to Edward Hibberd Johnson, December 1888

Editor's Notes

I am instructed by Mr Edison to submit to your notice the following report upon the experiments with the Edison Electrolytic Meter conducted at his laboratory in Orange N.J. and in accordance with your wishes expressed in your letter to him of the 20th August last.##This report is divided into three principal sections as follows:--##1st The general discussion of the electircal working parts in the existing forms of meter and the degrees of accuracy under ordinary conditions obtainable together with the principal source of error##2nd The means can be adopted to eliminate or reduce these errors in the existing form of meter and under working conditions.##3rd The improvements in the existing form capable of being made in accordance with the results of the entire investigation.##Section 1.##General features##The accuracy of the Edison Electrolytic meter so far as its electrical operation is concerned depends upon the maintenance under all practical conditions of a fixed ratio between the resistance of the two branches of a divided circuit generally known as the shunt circuit, and - the bottle circuit: together with the restruction of the counter E.M.F. of polarization in the bottle circuit to certain due limits.##To determine practically how closely the indications of the meter were to be depended upon, under ordinary conditions, a series of experimental trials was made on various leads. The complete description and detail of their trials is given in Appendix A.##The conclusions afforded by these trials -- :--##1st That when meters are worked steadily at above 50% of this full lead, their indications are generally reliable to 2 o 3%.##2nd That when meters are worked at above 5% of their full lead but with marked intermission or period of rest they tend to underindicate.##3rd That this tendency to underindicate [inacases?] as the load on ht emeter diminishes. In actual trial error of as much as 15% have been met with at 20% & full load and under very unfavorable conditions there are other experimental reasons for believing that a meter might be capable of underindicating as much as 25%.##4th That this tendency to underindicate which##5th That the temperature compensation does not appear to be complete, the underindication being greater at lower temperatures.##So that generally speaking the results go to show that a meter indicates satisfactorily when in operated continuously on above 50% of full load, but that its indications fall short of the mark as the current continuity, the lead, and the temperature diminish.##Since the majority of meters work discontinuous being idle at last 12 hours in 24, a deficit error in practice is to be expected depending upon the individual lead of each meter.##[dividing line]##Disenssion of the variation in resistances.##(1) The shunt circuit.##The shunt being made of german silver strip is after being once calibrated and fixed, only liable to temperature variation in resistance provided that no overheating at any time recurs. This temperature variation being about 1/20th of one per cent per degree centigrade is probably confined in practice to a range of 1% above and below the normal, and this is therefore the probable limit of error in this indication of the instrument due to the shunt.##The calibration of the shunts sums from the meter that have been tested, to be close and reliable. A comparison of 10 shunts in 5 meters is given in Appendix B and shows that they all agreed to about one quarter of one per cent, their obseved difference being in fact probably as much due to the measurements so to the shunts.##[dividing line]##The continuous operation of a No 4 meter for many hours on heavy loads (above 10 amperes) seems to generally produce irregular loose deposit commonly known as ['shunt?' 'smut?'] on the cathode, which in time will bridge over that liquid layer between the plates and shortcircuit them. The above statement requires therefore to be taken with this reservation##(2) Resistance of the bottle circuit.##The resistance of the bottle circuit is composed of three parts##(1) The metallic resistance, including the copper wire on the spool, the clips, rods, and zinc plates.##(2) The liquid resistance of the zinc sulphate solution in the bottle.##(3) The transition resistance resident in the bottle at the bounding surfaces between plates and solution.##Since the resistance of the shunt circuit may be regarded as invariable, the entire electrical accuracy of the meter depends upon the constancy of the sum of these three resistances under all practical condition, as well as the absence of under counter-electromotive force.##The metallic resistance, if ordinary care is taken in the operation of the elipcontacts, is subject only to temperature variation.##The liquid resistance of the zinc sulphate solution depends upon its density, purity, temperature, and [spage?] dimensions as well as on the extent of the metallic surface it covers.##The transition resistance depends only on the chemical or mechanical condition of the immersed surface of zinc, and is thus affected by the electrolyte changes set up during the action of the meter, by the strength of the current, its continuity, &c.##It was soon found experimentally, that while the metallic and liquid resistance area practically only affected by temperature in any given bottle and circuit, and thus readily under control; the transition resistance was irregular, sometimes large, very erratic in behavior, and in fact was the great source of imperfection.##Transition Resistance##Effect of amalgamation-##A member of measurements appear in Appendix C made with a view to determining the amount of transition resistance under different conditions. The results appearing on page - show that the transition resistance for two pairs of No. 16 plates carefully cleansed and immersed was about 15 for one and 8 for the other on first measurement. Throughout all the subsequent tests with these plates, the transition resistance was kept down to about 0.18 ohm at ordinary temperature by the use of [mercury?] in amalgamation.##This high transition resistance can be accounted for by the supposition favoured in other evidence, that a thin layer of air closely envelope the metla plates adhering to them after immersion, thus insulating them more or less from the surrounding liquid. The mercurial amalgam seems to displace this gaseous envelope and to act as an efficient intermediary conductor being itself semiliquid. Even with freshly amalgamated plates however, the results shew that ther is a distinct transition resistance varying with the temperature, being about 8 times greater at 5" than at 50 [degree symbol] C. If the meter be then set to work so if a current be passed through the bottle, this resistance tends to diminish as though the electrolytic change of surface promoted a still more intimate liquid contact. In a few hours however other purely chemical changes set in at the metal surfaces, which results in a speedy increase of transition resistance.##Observations of fire six light meters working through 40 ---fers in the day time##[chart follows]
Supplied year and month




Folder/Volume ID


Microfilm ID


Document ID



Thomas A. Edison Papers, School of Arts and Sciences, Rutgers University
Download CSV | JSON