Measurements are reported on carbon resistors and some related types (frequency range 2 c/s --1.2 mc/s) and on thermistors (2 c/s --50 kc/s). For carbon resistors we found that the noise spectrum in the whole range can be represented by S(v) = Ai2/v a, with a ~ 1, where i is the current and v the frequency. This is in contrast to measurements of Campbell and ChipmanS) and not in accordance with theories of Macfarlane and Petritz xs) l~). Measurements on thermistors showed that the current noise is extremely weak as compared with carbon resistors. The spectrum showed deviations from the I/v dependence. The spectral density in the "l/v part" of the spectrum is proportional to i 1'2. This is in agreement with measurements of B r o p h y n), published recently. Some explanations are suggested for the other part of this spectrum.
Introduction. For the spectral density of thermal voltage fluctuations in a resistor, N y q u i s t derived the formula
where k is Boltzmann's constant, T is the temperature, R is the resistance and the index th refers to the name thermal noise. This formula is strictly valid only if no current flows through the resistor but it also holds fairly well for current carrying wirewound resistors. In 1932 Williams and Thatc h e r 1) reported that the noise in current carrying carbon resistors is much larger than according to (1). They did not pay further attention to this fact. B e r n a m o n t 2) discovered the same effect in metallous films through which a current flowed. S u r d i n 3) extended these measurements. Other authors called the noise in excess of thermal noise, given by (1), c u r r e n t n o i s e. For the spectral density of current noise, B e r n a m o n t found Sv.(V ) dv = Ai 2 dv/v ' ~ (0.88 ~ a < 1.1)
(2)
Afterwards many investigations were made about this current noise in conducting media. We mention the experiments on carbon resistors and *) In our notation the first index ( V or i) refers to mean square voltage fluctuations, resp. to mean square current fluctuations, the second one to the kind of noise, we are dealing with.