The Early History of Seismometry (to 1900)

James Dewey and Perry Byerly

The Tilt Controversy

For a seismograph to be effective, its response characteristics must be known. We have seen that the British seismologists in Japan attempted to simplify the problem of the seismograph response by building relatively long-period instruments, and assuming that the seismogram trace was proportional to the ground displacement. The general problem of the response of a damped oscillatory system to ground motion of arbitrary period was considered theoretically by Perry and Ayrton (1879). Poincaré (1888) and Lippmann (1890) presented short notes on the integration of a seismogram trace to obtain ground displacement.

In the 1890's, however, the possibility arose that all of the theories of seismograph response just mentioned were fundamentally wrong. For these theories assumed that a seismograph pendulum responds to linear motion which is in the plane of oscillation of the pendulum and perpendicular to the line joining the pendulum's center-of-mass with the pendulum's axis-of-rotation. That is, a common pendulum which oscillated in an east-west plane was expected to respond to linear horizontal motions occurring in the east-west plane. The alternative theory, which became very widespread in the 1890's, was that horizontal and common pendulums were responding largely to tilting, which was in turn due to vertical displacements of the Earth's surface. As the Earth's surface tilted up and down, the pendulum would attempt always to point in the direction of gravity (or to point along the projection on the pendulum's plane-of-oscillation of the direction of gravity).

At first, the implications of recording tilting instead of displacements do not seem to have been understood among seismologists in Europe. We shall see, however, that calculated ground displacements commonly varied by several orders of magnitude according as the seismograph was assumed to be responding to horizontal displacements or to tilts due to vertical displacements. To be confident of any seismographic measurements of the amplitude of ground motion in an earthquake, it was necessary to determine the importance of tilting in earthquake motion.

An instrument for recording tilt in an earthquake was suggested as early as 1703 (De la Haute Feuille, 1703), but the idea that seismographs should record tilting in earthquakes was not widely accepted until European seismologists began recording teleseisms. Then, it was assumed without much argument, at first, that the recorded waves represented tilt, rather than horizontal motions (Ehlert, 1897a; Schlüter, 1903, p. 301-325). Perhaps this was because the earliest records of a teleseism, those of von Rebeur, were obtained with instruments which were built to measure changes in the direction of the vertical. Furthermore, the recorded motion, consisting of long-period sinusoidal oscillations, suggested a wave motion not unlike waves on the surface of the sea, which seemed to imply considerable tilting (Agamennone, 1894).

While the European seismologists were interpreting their first seismograms of teleseisms, John Milne, still in Japan, had independently concluded that tilting had a significant effect on the response of a seismograph in an earthquake. During the Mino-Owari earthquake of October 28, 1891, Milne (1893d) had observed horizontal-pendulum seismometers located at a distance of 140 miles from the epicenter and was convinced from watching the pendulums that the instruments were being tilted. A comparison of seismograms from different horizontal instruments suggested that record amplitudes were proportional to the static tilt sensitivities of the instruments rather than to their static displacement sensitivities. In order to obtain a more accurate measure of tilt, Milne (1893e) set up a beam balance, arranged so that the motion of the balance's vertical pointer was amplified and recorded on a smoked-glass plate. Milne believed that the beam would remain horizontal in an earthquake while the Earth tilted beneath it. The balance was in stable equilibrium, so that one would expect it to respond also to horizontal displacements of the ground, like any other pendulum. Nevertheless, Milne seemed to assume that the records given by the instrument represented only tilt. Significant "tilting" was recorded by the balance seismograph. Milne was forced to the "unpleasant conclusion ... that all the records hitherto published in Japan where vertical motion has been recorded are of but little value" (Milne, 1893e, p. 103-104).

Figure 25. Schlüter's klinograph (modifed from Schlüter, 1903). The frame rotates on an agate edge S, which rests on an agate plate, not shown. S is made to coincide with the frame's center of mass by adjusting the weights L. Light is reflected off the mirror M onto a moving photographic surface. The mirror is mounted so as to apply a slight restoring force to the klinograph.

Soon, however, the "tilting hypothesis" met objections. A major difficulty was the large vertical displacement required to produce a tilting of the magnitude which seismologists believed they were recording. Under the assumption that his seismographs were recording only tilt, Cancani (1894) calculated that the ground rose and fell forty centimeters during the passage of waves of sixteen-second period from a distant earthquake. The improbability of such large vertical motion occurring in the unfelt waves of a distant earthquake was emphasized by Schmidt (1896). The same record which Cancani thought represented tilting due to a vertical displacement of forty centimeters, Schmidt pointed out, could also represent a horizontal displacement of much smaller amplitude. (In this case, using Schmidt's formulas, the horizontal amplitude would be less than a millimeter.) Feeling that a horizontal displacement of a millimeter was more probable than a vertical displacement of forty centimeters, Schmidt concluded that the common and horizontal pendulums were indeed recording horizontal oscillations from teleseisms.

Schmidt's conclusion was not accepted by many seismologists, at first, and a controversy arose over the presence or absence of large tilts in earthquake waves. (Ehlert (1897a) summarizes the arguments which were used "for" and "against" tilting.) Although the large vertical displacements required to produce significant tilting were not perceptible to human beings, Ehlert (1897a) argued that the vertical displacements might still exist, but that their periods were too long to be noticed by humans. At this time, there were no long period vertical seismographs, and seismologists do not seem to have considered the possibilities of directly recording the vertical displacements believed to be responsible for tilting.

In 1899, W. Schlüter (1903) began recording in Göttingen with a tilt-measuring device, which he called a "klinograph". The instrument was similar to a beam balance, with the difference that the horizontal axis of rotation of the moving frame, or beam, passed through the center of mass of the frame (Figure 25). In this situation, the frame would not respond to linear displacements. By virtue of its rotational inertia, however, the frame would respond to a rotational motion in a vertical plane, i.e., tilting. In order to stabilize the klinograph, Schlüter adding a restoring force to one end of the beam. The instrument recorded photographically.

Schlüter operated his klinograph simultaneously with a horizontal pendulum seismograph. He calculated the tilt sensitivities of the two instruments and found them to be of the same order of magnitude. If an earthquake were recorded on the horizontal pendulum seismograph, and if the seismogram from this instrument represented tilting of the ground, then the earthquake would be well recorded on the klinograph. Contrary to what he had expected, Schlüter found that the klinograph recorded nothing at times when the horizontal pendulum seismograph recorded large earthquakes. He concluded that the horizontal pendulum was responding only to horizontal displacements, for which the klinograph had no sensitivity. Tilting, if it existed, was too small to affect either his klinograph or contemporary horizontal pendulum seismometers.

Schlüter's experiment, and theoretical considerations such as those of Schmidt, were accepted by many as strong evidence in favor of the viewpoint that seismographs responded largely to linear motion rather than tilting, except for waves of very long period (Wiechert, 1903). The experiment provided justification for neglecting the effect of tilt in theoretical studies of seismograph response. Those (such as Galitzin, 1902) who doubted the validity of Schlüter's experiment found the theory of a pendulum responding to both tilt and displacements to be so complicated that they were forced to neglect the effect of tilt as a matter of convenience (Galitzin, 1904). [However one of the authors of this paper (P B.), remembers when Father Macelwane changed the interpretation charts for the Berkeley Bosch-Omori pendulums to correspond to response to tilt rather than displacement. This was in 1923.]

From the Bulletin of the Seismological Society of America. Vol. 59, No. 1, pp. 183-227. February, 1969.