In different problems in physical and electrical measurements, one of the most serious problems is that of securing reliable and frequent time‐signals. While the rates of good clocks and chronometers can be determined from the Arlington signals as supplied by the Naval Observatory to the required accuracy, the second or half‐second signals obtained from them are not sufficiently reliable and in some cases not of sufficient frequency to meet the needs. For some of these problems, what is needed is 20 and preferably 100 or 1,000 signals per second uniformly spaced over each second to a 0.0001 second or, better, and uniform over periods of an hour or mora to a part in a million or better. Second‐to‐second tests of the signals obtained from the Bureau of Standards Riefler clock through the contact supplied with it and relays show a dispersion of 0.01 second and more, so it would not meet the requirements even if the signals were of sufficient frequency. Figure 1 shows a record of the signals from the Riefler clock as well as from a chronometer. The missing signals are at the 59th seconds of alternate minutes. The more uniform signals are from the chronometer, and these serve to show that the scattering of the signals from the Riefler clock are not to be attributed to the recording‐