Some of the prevailing origins of the noise in the period range of several ten seconds were investigated using the Press-Ewing seismograph system at the Abuyama Seismological Observatory. A simultaneous observation of noises with seismometer, microbarograph and highly sensitive resistance thermometer was carried out in order to estimate the effect of the barometric or temperature variation on the noise. The vibration recorded on the vertical seismograms with a magnification of 800 at 15sec has a good correlation with the barometric pressure change; thus it is thought that this noise may be affected by the buoyant force of the pendulum of the vertical seismometer. The noise therefore can be almost completely eliminated by enclosing the seismometer in a pressure-tight vessel. on the other hand, the noises on the horizontal seismograms have no correlation with the microbarograms, so that these noises may be mainly caused by local inclination of the ground.
Temperature variations seem to have little effect on the Press-Ewing type seismometers. Contrary to this, very high noise appears on the Galitzin type seismometers throughout the winter, or whenever the underground temperature is higher than the room temperature. A crude analysis on the fluid mechanics shows that the turbulence of air occurs inside a large case such as a Galitzin type seismometer. In order to prevent this turbulence, the size of the case should be made as small as that of a Press-Ewing type seismometer and/or the temperature distribution inside the case should be kept warm in the upper portion and cold at the bottom.
Any noise caused by a barometric pressure change, turbulence of air and the inclination of the ground must have originated by the forces which act on the pendulum directly. In this experiment, the maximum sensitivity of the vertical seismograph against the barometric change, and that of the horizontal one against the ground inclination are estimated as 9cm/mb and 0.006"/cm at 150sec, respectively.