If x-ray tubes gave monochromatic radiation, specification of quality could be given equally well as kilovolts, half-value layer (h.v.l.), or absorption coefficient (μ in any named material). But since the radiations we use clinically are heterogeneous, we feel it necessary to give all the factors which influence the wavelength composition (the spectrum) significantly, namely the voltage and the filter, as well as the h.v.l. In the pioneer days, the only convenient way to measure high voltages was by spark-gap, which for pulsating voltages measures the peak voltage (kvp), not the effective voltage. In the early days the quantity of radiation was hard to measure, too, and it was customary to infer the quantity from the tube voltage, current, and time. Since output increases very rapidly with voltage, and since penetration also increases very rapidly with voltage in the range available in those days, this factor came to have a dominating position in the radiologist's thinking. Nowadays we have good means of measuring quantity and quality of x-rays, but our biased attitude toward voltage persists. We now know that comparatively gross differences in quality make but small differences in clinical application. Nevertheless, in working out new roentgenographic technics and in checking calibrations of machines, precise knowledge of the kilovoltage is very desirable. Modern shock-proof apparatus makes it inconvenient to bring out the high tension for measurement by sphere-gap. The radiologist must accept the manufacturer's calibration of kilovolts peak against line voltage, with tabular (or built-in) corrections for the several tube loadings. He is naturally skeptical of these calibrations, because the same exposure at the same kilovolt setting fails to give the same density in the roentgenogram on two different machines. It is possible to infer the voltage on the tube from the hardness of the radiation. If the voltage is pulsating, the inference is not from the peak voltage, but from the effective voltage. The greater the thickness and the higher the atomic number of the filter used for a given kilovoltage, the harder the transmitted radiation will be and the nearer the effective voltage will approach the peak voltage. It is, of course, really the hardness itself that is of clinical significance. But our thinking has been done all these years in terms of kilovolts as the measure. We are not here defending this illogical attitude. We are proposing, rather, to yield to the prejudice and describe an instrument to measure the hardness and give the reading in the terms requested, namely kilovolts. Construction of the Kilovoltmeter The inferential kilovoltmeter (Fig. 1) consists of two ionization chambers, each of which is surrounded by its own filter. The larger ionization chamber is twenty times as sensitive as the smaller one.