## Extract

The measurement of alternating currents of very high frequency in general presents great difficulty because all existing ammeters have a frequency correction which cannot be calculated. The dynamometer ammeter of ordinary construction is very satisfactory for currents whose frequency does not exceed a few hundred cycles per second, but is unsuitable for currents whose frequency is many kilocycles because the highly inductive winding presents an enormous impedance to the current and its presence in a circuit modifies completely the conditions obtaining therein. Further, the distribution of current over the cross section of the wire and the value of the current from turn to turn will alter with the frequency, with the result that a steady current calibration becomes invalid and the correcting factor cannot be calculated or even estimated roughly. So even if its presence can be tolerated in a circuit, such an instrument can be used only for relative measurements at one frequency and its indications cannot be reduced to absolute measure. The ammeters in general use are thermal instruments depending on the thermal expansion of a suitable element or on the production of a thermoelectric E. M. F. in a junction placed close to a wire heated by the high frequency current. In either system the resistance of the heated element depends on the frequency of the current which heats it and so a steady current calibration cannot be used indiscriminately. The necessary configuration of the heated element and its situation with respect to surrounding alternating magnetic fields usually renders impossible the calculation of its resistance. It is usual to make the heated wire of a high resistance material and with a small diameter so as to render the calibration sensibly constant up to a high value of frequency. If the calculated resistance of such a straight isolated wire at an assigned frequency has increased by only a small fraction of 1 per cent, above the steady current value, then it will seem reasonable to suppose that the calibration of the bent and unisolated heated wire is valid to at least 1 per cent, up to this frequency. For higher frequencies there will be a correction term whose value can be estimated only very roughly. To maintain the calibration, valid np to a frequency of some thousand kilocycles per second the heating wire must be so fine that it will not carry a current of more than, say, 1 ampere. To measure larger currents we are faced by the problem of providing a shunt which the shunting ratio is independent of frequency. Brief consideration will show it is very difficult to arrange a group of fine parallel wires so that each has precisely the same resistance, and further that the situation of each one is the same with respect to all the others and also the remainder of the circuit. If both conditions are not fulfilled the total current will not always divide equally among the component parallel paths and the calibration curve will be subject to a frequency correction. A common and successful method is to arrange the parallel wires as generators of a cylinder, double cone, or hyperboloid and to allow each wire to heat one of a group of thermocouples connected in series electrically.

## Footnotes

This text was harvested from a scanned image of the original document using optical character recognition (OCR) software. As such, it may contain errors. Please contact the Royal Society if you find an error you would like to see corrected. Mathematical notations produced through Infty OCR.

- Received May 21, 1928.

- Scanned images copyright © 2017, Royal Society