Popular Science Monthly/Volume 69/August 1906/Seismograph and Magnetograph Records of the San Francisco Earthquake, April 18, 1906
| SEISMOGRAPH AND MAGNETOGRAPH RECORDS OF THE SAN FRANCISCO EARTHQUAKE, APRIL 18, 19061 |
By Dr. L. A. BAUER
DIVISION OF TERRESTRIAL MAGNETISM, U. S. COAST AND GEODETIC SURVEY.
THE San Francisco earthquake was one of several large earthquakes recorded the world over since the beginning of this year. The writer's prime interest in it as a magnetician is in the record it left behind on the magnetographs at various magnetic observatories of the United States Coast and Geodetic Survey.
It has happened several times within the last few years that earthquakes have occurred in this country which were not recorded for one reason or another, on the existing seismographs, but were indicated by the record of certain magnetographs. The most notable instance was the New England earthquake of March 21, 1904, at about eight minutes after one o'clock in the morning, eastern time. Seismographs of the Milne type at Toronto, Canada, and Baltimore, Maryland, and of the Bosch-Omori type at the Weather Bureau, Washington, D. C, failed to give any record of this earthquake, which was appreciably felt throughout the New England States. The magnetograph at the observatory, Cheltenham, Maryland, sixteen miles southeast of Washington, gave a distinct record at 1h 05s to 1h 17m eastern time. So there have been a number of earthquakes recorded by the magnetograph at Baldwin, Kansas, which were felt in the middle states and reported in the papers. In fact, at this observatory, situated in a region where felt and unfelt local and regional earthquakes are comparatively frequent—note for example the many recent occurrences—more records of earthquakes are obtained on the magnetograph than at any of the other magnetic observatories.
This repeatedly authenticated fact made desirable a concurrent study of seismograph and magnetograph records and hence seismographs have been installed within the last two years at all of the magnetic observatories excepting at Baldwin, Kansas, which was omitted because of its probable early removal on account of the possibility of disturbing influences from electric-car lines. So it happens that the Coast and Geodetic Survey is able at present to contribute the principal portion of the accurate observational data of earthquakes obtained in this country. It was with the expectation that magnetic observatories would also be excellent stations for the installation of instruments that the writer, while attending the Seventh International Geographic Congress at Berlin, 1899, as a delegate from the National Geographic Society, was made a member of the Provisional Committee of the International Seismological Association, just organized by the congress.
The instrumental seismological data referring to the recent San Francisco earthquake will be contributed from the following stations in Canada and the United States:
Table 1. List of Stations and Institutions in Canada and the United States Contributing Seismological Data.[1]
The exceedingly sparse distribution of seismological stations in this country is made apparent by this list, there being none in the middle portion of the United States, where, as already stated, regional earthquakes are comparatively frequent. It is therefore fortunate in the study of the San Francisco earthquake that we may have recourse also to the data afforded by magnetographs, especially by those at Baldwin, Kansas, and Sitka, Alaska—the nearest magnetic observatories to the origin of the quake and situated, as will be seen from Table 3, at about the same distance from San Francisco. So also is it a fortunate circumstance that we have both magnetograph and seismograph data from the two magnetic observatories, Honolulu and Cheltenham, which are also practically equidistant from the origin.
Now a peculiar circumstance is that this earthquake, while giving a record on the seismograph at the Porto Rico Magnetic Observatory so large as not to be fully recorded, left no trace behind on a magnetograph of the very same pattern as at the other observatories. On the other hand, the equally large earthquake of January 31, last, the origin of which was at sea off the west coast of Ecuador, besides recording itself on seismographs the world over was recorded on the magnetographs at Baldwin, Porto Rico and Cheltenham, but this time not at Honolulu. This seaquake was accompanied by a tidal wave twenty feet high which rushed in on the coast of Ecuador, causing great devastation; it set the Pacific Ocean in vibration, which according to the tide-gauge records of the Coast and Geodetic Survey at San Diego and Honolulu lasted for three days. The tidal wave, when it rushed in on the Hawaiian coasts, was several feet high, and the record of this quake of January 31, as recorded on the Milne seismograph at the Honolulu Magnetic Observatory, was among the largest since the installation of the instrument, September, 1903, and yet the delicately suspended magnets, as far as the magnetic records at this observatory would indicate, were not affected.
Why is it that an earthquake will at times be recorded by magnetic instruments and at other times leave no record? Or, to go back to the fundamental question, what do the magnetic instruments record—an actual mechanical effect due to the mechanical vibration of the point of support? If the observed effect is a purely mechanical one, then why is it that not every mechanical disturbance is recorded on the photographic records of the fluctuations of the magnetic needles? What is the characteristic of the mechanical vibration, the presence or absence of which in the earth movements is responsible for the presence or absence of the effect recorded by magnetic needles?
The solution of these questions may show the magnetograph to be a most useful adjunct to the present instrumental equipment for recording earth movements.
Is the possibility of any actual magnetic effect accompanying an earthquake entirely to be excluded? If so, in the case of the distant earthquakes, as seems probable, is the possibility also to be excluded for the less distant ones, or say for stations within a certain prescribed region about the origin of the quake? Are those cases where records are secured on magnetographs and not on seismographs to be attributed possibly to such a magnetic effect which has no influence on instruments responding merely to mechanical vibration? Or is it possible that the magnetograph is in certain cases a better micro-seismograph than the Milne or Bosch-Omori instruments used in this country?
We have thus some extremely interesting questions presented to us which, however they may be solved, will be a valuable contribution to our knowledge of earth movements. The possibility might also be mentioned that an approaching earthquake might through electric or magnetic effects give the first indication on magnetographs because of the much greater velocity of propagation of such effects than that of the mechanical vibrations. We know that a magnet subjected to strain undergoes changes in its magnetization and so the question arises whether the earth's magnetized rocks may not likewise give some indication of their state of strain during an earthquake by slight magnetic fluctuations. Or, an earthquake may be accompanied by a redistribution of the magnetic rocks or of the electric earth-currents known to exist, and thus give rise to a possible magnetic effect.
Enough has been said to show that a careful and exhaustive investigation of seismic effects recorded on magnetographs is certainly one that merits undertaking. The Department of Terrestrial Magnetism of the Carnegie Institution of Washington, in connection with the study of the magnetic effect, recorded simultaneously over the entire globe coincident with the outbreak of Mt. Pelé, on May 8, 1902, is making a systematic study of the volcanic and seismic effects recorded by magnetic instruments with the.cooperation of the Coast and Geodetic Survey and of the Canadian Meteorological Service. A paper by Mr. J. E. Burbank, published in Vol. X., p. 113, of the journal, Terrestrial Magnetism, brought the investigation up to the time of the installation of the seismographs at the Coast and Geodetic Survey Observatories two to three years ago; a second paper, to be published in the course of the year, will continue the research as based upon seismic and magnetic instruments in operation at the same observatory.
It had been noticed for some time that magnetic instruments responded to certain earthquakes, but the cases noted were of such a class as to convey the first impression at once that the effects recorded were mechanical ones. Milne in 1898 made quite an exhaustive investigation of this class of effects for the whole earth and covering the period from 1889 to 1897. He likewise found that these effects were not invariably recorded at every magnetic observatory. He considered the results inconclusive and deemed it necessary to await the time when both seismograph and magnetograph records could be had at the same place. A recent notable contribution to the subject based on magnetic records at one observatory, without, however, at the same time corresponding-seismological data, has been made by Dr. Messerschmitt, in charge of the Munich Magnetic Observatory.
Previous magneticians, such as Eschenhagen, Wild and Liznar, had found that from a comparison of the effects recorded on magnetographs at various European observatories the effects, in certain notable earthquakes, progressed from station to station with the velocity of about three kilometers i. e. the rate of propagation of the long or surface seismic waves. This measurable difference in time between any two stations and its correspondence with the time interval required for the transmission of the surface waves was a very good indication that a purely mechanical effect had been recorded and not a distant magnetic effect, as the latter would have been observed simultaneously at all stations. Or, if it was a magnetic effect, then in each case it was due to one of the possible local causes enumerated brought into action upon the arrival of the mechanical disturbance at the particular station.
In our study, however, it has been possible to differentiate much more closely and at times to separate the effects on the magnetic records into the various phases—preliminary tremors and principal portions, etc.—in a manner analogous to usual treatment of the seismograph records. A notable instance was the destructive Guatemalan earthquake of April 19, 1902, which, as may be recalled, preceded the Antillean volcanic eruptions of that period. At that time there were no seismographs at the Coast and Geodetic Survey Magnetic Observatories; however, an inspection of the table below will show that with the records obtained on the magnetographs at Cheltenham, Baldwin, Sitka and Honolulu (the Porto Rico Observatory did not then exist), it is possible to study the seismic effects on them—even down to the preliminary tremors—equally as well as on the seismic records obtained at Baltimore, Toronto and Victoria. The earliest notice of this earthquake was received at Baldwin, the nearest station to the origin—Guatemala. Here then we have a notable case where the magnets were affected by even the preliminary seismic tremors, this being a different, case from the European ones cited above, as these tremors travel with a velocity of about nine kilometers or more per second.
There have been many other similar instances and it has even occurred at times that the magnetic instruments have given a slightly
Table 2. Seismograph and Magnetograph Records of the Guatemala Earthquake, April 18, 1902.
earlier record than the seismograph. The effect is such a characteristic one that when it has once been recognized it will not be mistaken for any ordinary magnetic effect. Four types can be distinguished: First, those in which the disturbance begins abruptly and quickly reaches a maximum, dying down gradually (usually the case in a near-by earthquake); second, those in which a small preliminary effect precedes the principal portion, and in which there are often two or more maxima; third—by far the most common—those consisting of a small diamond-shaped' disturbance; and fourth, those in which the trace is simply blurred and broadened.
With these introductory statements as to the possible relation between seismology and terrestrial magnetism, let us now pass to the consideration of the recent San Francisco earthquake.
Table 3 contains the results of the records obtained up to date (May 17) at the office of the Coast and Geodetic Survey, both from magnetographs and from seismographs. It will be seen that the region embraced extends from Honolulu on the west to Hungary on the east, or about one third the way around the globe. All necessary data, such as latitude, longitude, distances from San Francisco along the surface, as well as along chord, chord depth, etc., etc., are found in the table.
Next are given the Greenwich mean times (0 to 24 hours, midnight to midnight) of the occurrence of the various phases of the seismic disturbance as recorded on the seismographs. For the preliminary tremors, phase I (longitudinal waves), next phase II (transverse waves), then principal portion (surface waves), etc., etc. It is particularly interesting to compare the times for Cheltenham, Washington and Baltimore and to note how closely they agree. Owing to the slight difference in distance of these three different stations from San Francisco the times should not of course be strictly the same, though the difference should not be more than a few seconds. Considering the totally different types of instrument (Milne at Baltimore and Bosch-Omori at Washington and Cheltenham), certainly the comparison is very satisfactory.
It will be seen that the preliminary tremors were recorded by the seismographs at Honolulu and Cheltenham at about the same time, these two stations being at about the same distance from San Francisco. The reader will follow without assistance the progression of the various waves from station to station as given in the table. [Since this table was prepared many more records have been obtained which are of interest, notably the seismograph and the magnetograph records from the Toronto Magnetic Observatory.]
In Fig. 1 we have a reproduction of the seismograph record obtained at the Cheltenham Magnetic Observatory. The recording cylinder of the Bosch-Omori seismograph, around which is wrapped the 
smoked sheet of paper on which the record is made, makes one complete revolution in an hour, each sheet whether of the N.-S. or of the E.-W. component containing a whole day's record (24 lines). The distance between two of the dots represents one minute. In order to get a convenient size for the figure, it was necessary to omit about one third of the total length of the sheet, so that the distance from S. to S. or E. to E. represents about two thirds of an hour. On the original, the pointer or the recording stylus multiplies the motion ten times, hence in the reproduction the magnification is about three times. The maximum amplitude of motion was not recorded, the pointer striking the damping brushes. In deducing the actual displacement of an earth-particle at Cheltenham, it is necessary to take into account the period of the pendulum which for the N.-S. component was about 25 seconds and for the E.-W. component about 20 seconds and the period of the recorded earth-movement about 2 to-f seconds for the preliminary tremors and about 10 to 20 seconds for the principal portion. A rough calculation would give the total recorded horizontal displacement of the earth-particle, back and forth, of about 1/5 of an inch, which on account of the comparatively long period would not be felt by the human being. These explanations will doubtless be sufficient for the elucidation of the figure; for a description of the instrument the reader is referred to Dutton: 'Earthquakes.'
Passing next to the times recorded by the magnetograph (D stands for declination, H for horizontal intensity and Z for vertical intensity), it is seen that the effect in this instance did not begin at the four observatories where a record was obtained—Honolulu, Sitka, Baldwin and Cheltenham—until the arrival of the principal portion (long or surface waves) recorded on the seismographs, and that for this phase the agreement between seismograph and magnetograph is most satisfactory. It will also be noted that the time at Baldwin is intermediate between San Francisco and Cheltenham, so that the record of this observatory is a most desirable acquisition. Note also that the time is nearly the same as at Sitka, Baldwin being just a trifle farther from San Francisco than Sitka.
Next are found in the table the velocities of the various transverse waves—longitudinal, transverse and surface—computed along the paths indicated in the column on the extreme right. For the region embraced it will be seen that the longitudinal waves, which were the first to arrive, traveled at an average velocity of six miles per second, the transverse waves at an average velocity of 31⁄2 miles, whereas the surface waves had a velocity of about 2 1/3 miles per second according to seismograph and magnetograph. It takes about 3 hours and 20 minutes for these waves to pass around the earth completely, whereas the preliminary tremors, phase I (longitudinal waves) reach 
Fig. 1. Record of the San Francisco Earthquake as obtained on the Bosch-Omori Seismograph at the Magnetic Observatory, Cheltenham, Maryland, reduced three and one half times. [The records of the N-S component and of the E-W component were re arranged so as to bring the points of beginning of the preliminary tremors, phase I (A) in the same vertical line. B marks the beginning of the preliminary tremors, phase II and C the beginning of the long surface waves.]
Fig. 2. Record of the San Francisco Earthquake on the Eschenhagen Magnetograph at the Magnetic Observatory, Cheltenham, Maryland, reduced 2 times. [The hours as marked are approximately local mean time. The earthquake effect will be noticed on the three magnetic elements, horizontal intensity, declination and vertical intensity in the shape of a trumpet formation between 8 and 9 a. m., local mean time. The range or double amplitude of the disturbance was about 1,1000 part of the horizontal intensity and about 1/3000 part of the vertical intensity. On account of the intersecting of the curves, the range in the magnetic declination cannot be given.]
time of the shocks on the Pacific Coast which gave rise to the effects recorded at distant places as occurring at 5h12m Pacific time or 13h 12m, April 18, Greenwich time. There may have been earlier preliminary shocks.[2]
The amplification of this table to embrace the entire earth will be left to Professor Harry F. Reid, a member of the San Francisco Earthquake Commission.
Why is it that in this severe earthquake the magnets responded only to the long or surface waves and not to the preliminary tremors, and why did the magnets at Porto Rico give no record at all? These are the questions which I believe to be of concern not alone to the magnetician, but also to the seismologist and to the student of geophysics in general. Of the many earthquake records already obtained, there are a large number where the disturbance on the seismograph was considerably smaller than the San Francisco one and yet the magnetograph responded to even the preliminary effects. Evidently we must be getting a record of something on the magnetograph, not immediately evident from the present seismograph records, which causes this peculiar differentiation of seismic disturbances into the following classes: (a) recorded by seismograph and not by the most delicate magnetograph, (b) recorded by magnetograph and not recorded by seismograph, (c) recorded by seismographs and magnetographs partially (surface waves), (d) recorded completely by seismographs and magnetographs.
My present belief is that the effects recorded by suspended magnets are chiefly mechanical ones due to the vibrating motion of the points of support, though the possibility of a magnetic effect within a certain prescribed region of the origin of the earthquake, brought about as above explained, is not to be excluded. It is a notable fact that at the Baldwin Magnetic Observatory, where, as stated, so many seismic effects are being detected which are to be associated with the comparatively local earthquakes in the Middle States and which fail to make any record on seismographs as far distant as Washington, the effects corresponding in time to lightning discharges have also been found which in many instances resemble very closely the seismic effects.
In the case of the San Francisco earthquake, however, there can apparently be no question that what was recorded by the magnetographs was a mechanical effect (see Fig. 2). It is a matter of interest to note
that at Baldwin there was a Blight actual magnetic effect 3-4 hours before the shock was felt at San Francisco, to which no counterpart has yet been found on the Cheltenham records, indicating that this effect observed at Baldwin was not a cosmic one, but was due to some local circumstances. To associate' it with the San Francisco earthquake is not at present warranted.
Owing to the optical arrangement of the magnetograph, in order to produce an effect which will be evident on the recording sheet, it is necessary to have a turning movement of the suspended magnet. Any parallel displacement of the magnet—sidewise or up and down—will give no observable effect, an actual turning or rotary movement of the magnet must take place and for this purpose a turning couple must in some manner be introduced. Such a couple is produced when the magnet is drawn out of its normal direction with the aid of a bit of iron which is then quickly removed; the earth immediately acts on the magnet with equal and opposite forces applied near the extremities. and after performing a number of vibrations about its mean position the magnet settles down and then takes up the course pursued before the artificial disturbance. The effect thus produced is very similar to some of the earthquake effects. Were an earthquake accompanied by the generation of magnetic forces, the explanation of the observed effects would thus be very simple.
When the seismic motion is such as to produce a tilting or rocking of the support, it can readily be shown that because the suspended mass is a magnet, a turning couple is brought into play by the earth's magnetism causing the rotary, vibratory motion of the magnet about its mean position. Were the suspended material a non-magnetic mass of sufficient weight, no such turning would take place, but the mass would act more or less as a 'steady point.' However, it is quite possible that with the very light magnets weighing but 1⁄2 gram, and short suspensions used, we may also have to deal with a form of pendulum seismograph, in which the period of the pendulum is sufficiently small as to more readily respond to certain micro-seismic motions than either type of instrument at present in use in this country.
It would seem therefore that seismologists might be assisted in the solution of some of the problems as to the precise character of the earth movements recorded on seismographs by a careful study of the seismic effects recorded on magnetographs, especially if the effects both in the horizontal and in the vertical plane be considered, and if furthermore the record be obtained on a more open time scale, so as to be comparable in this respect with the best seismograms.
Whether the San Francisco earthquake caused a change in the distribution of the earth's magnetism within the affected region is at present under investigation.
- ↑ A Bosch-Omori seismograph procured for this observatory was temporarily installed at Baltimore by Professor H. F. Reid for a comparative study with his Milne seismograph.
- ↑ Professor George Davidson, of the University of California, determined the time of first shock at his home in San Francisco by counting the number of seconds it took him upon awakening and going to his watch and noting the time. Owing to his large experience in the work of the U. S. Coast and Geodetic Survey, the time which he gives he deems to be correct within two seconds, via., 5: 12: 00 Pacific time. Professor A. O. Leuschner, of the University of California, according to his article in the Berkeley Reporter, of Berkeley, Gal., April 20, 1906, says: "The best record of the beginning of the heaviest shocks is furnished by the standard clock of the Student Observatory, which stopped at 5h 12m 38s Pacific standard time, while less severe shocks were recorded by Mr. S. Albrecht some 35 seconds earlier. The principal part of the earthquake came in two sections, the first series of vibrations lasting about 40 seconds. The vibration diminished considerably during the following 10 seconds and then continued with renewed vigor for about 25 seconds more. But even at this writing, about noon, the disturbance has not as yet subsided, as slight shocks are being recorded at frequent intervals on the Ewing seismograph, which has been restored to working order. [This seismograph was thrown out of action at the beginning of the earthquake; however, a fairly complete record was obtained with the duplex instrument.] The principal direction of motion was from south-southeast to north-northwest. The remarkable feature of this earthquake aside from its intensity was its rotary motion. The sum total of all displacements represents a very regular ellipse and some of the lines representing the earth's motion can be traced along the whole circumference." From this we deduce the time of the first shock 5:12:03.At the Lick Observatory the first shock was recorded at 5:12:12.At the Ukiah Latitude Observatory the first shock was recorded according to Dr. Townley at 5:13, correct within two or three seconds.At Eureka, California, the first shock as reported to Professor Davidson was noted on a regulator owned by H. H. Buhne, who was awake at the time at 5:11.As it is likely that the epicenter was somewhat west of San Francisco, but at no considerable distance, owing to absence of tidal waves, it is probable that the average time of the shocks at the origin which produced the records at the distant observatories was not far from 13h 12m, Greenwich mean time, which is at present adopted. L. A. B.