Groph!,.~.J . Int. (1997) 131, 421 422
SPECIAL SECTION-ASSESSMENT
OF SCHEMES F O R EARTHQUAKE PREDICTION
Forecasting earthquakes*?
J. B. Macelwane, S.J.
It was twelve minutes after five in Sail Francisco on the fateful
morning of April 18, 1906. Most of the city’s inhabitants were
still in bed when the earthquake struck and the great San
Francisco fire was started. Why was the public not warned of
the coming catastrophe? Why were the people of San Francisco
not told of the imminent danger of the earthquake so that
they could save themselves and their movable possessions?
Why were the city officials not alerted so that the gas and
electricity could be cut off? We may ask similar questions
about our other destructive earthquakes-about New Madrid
and Charleston, about Santa Barbara and Long Beach, about
Helena and El Centro.
Why d o we not have an earthquake-forecasting service to
parallel the United States Weather Bureau? Storms are forecast;
why not earthquakes? From time to time articles have appeared
in the newspapers about men who claimed to have arrived at
a complete solution of the problem of earthquake forecasting.
They give you long lists of supposed verifications to prove the
success of their predictions. Some forecasters base their predictions on the influence of the moon, others on the relative
position of certain planets, others on sunspot activity, and no
doubt there are persons entirely sincere who are allured and
deceived by a specious theory and by a series of apparent
verifications, but who lack sufficient scientific background to
make a critical appraisal of the factors involved.
No doubt, too, there are charlatans and notoriety seekers.
But seismology, the study of earthquakes, is a highly technical
science, and it must be pursued by the patient, objective,
fact-seeking method of scientific research.
If the earthquake prediction is made general enough, it will
not be difficult to find some earthquake that would seem to
verify it. Therefore it is essential, before we proceed further in
the discussion, that we clarify our ideas. We must be sure that
we are all talking about the same thing.
It is not enough to predict that earthquakes will continue
to occur in the future. To a person with any knowledge of
earth science, that is an obvious truth. Actually, more than
one million earthquakes occur every year. Would I be forecasting in any true sense if I told you an earthquake will occur
in Japan next week? Any experienced seismologist knows that,
on the average, two dozen earthquakes occur in Japan every
week. My prediction of an earthquake in Japan next week
would be a statement of a moral certainty; but it would be
too vague to be of any help as a forecast. What, then, is
a forecast?
*Text of a talk delivered in the course of the Philharmonic Symphony
broadcast over the CBS network, January 13, 1946. Manuscript
received from the sponsor, the United States Rubber Company, with
permission to publish, January 15, 1946.
t Reprinted from Bull. seism. SOC. Am., 1946, 36, 1-4.
0 1997 Seismological Society of America
A forecast, in the modern sense, is not a mere guess. Neither
is it a prophecy. A forecast is a definite statement in regard to
the sequence of future events, based on a balancing of probabilities derived from experience, observation, and the laws
which are known to govern such happenings. I think we can
agree, from the analogy of weather forecasting, that an earthquake prediction would not be called an earthquake forecast
unless it were sufficiently probable and specific to serve a
useful purpose. A public prediction of an earthquake will be
useful if it serves to protect life and property without arousing
undue panic.
An earthquake forecast must be spec@ This means that
the forecaster must predict three things about the earthquake.
In the first place, the forecast must announce the time of the
earthquake, at least within a few hours. Secondly, the forecast
must specify the intensity of the coming earthquake, or the
extent of the damage to be expected from it. Thirdly, the
forecast must state in just what place or places the destruction
of property is going to occur.
Besides being specific, the forecast must also be reliable.
That is, a predicted disaster must be sufficiently probable to
justify public authorities in removing the threatened populations, cutting off the gas supply and the electricity, and
otherwise disrupting the normal life of the people. The mere
heralding of a disastrous earthquake would create a panic and
might drive many people insane.
1s it possible, in the present state of scientific knowledge, to
predict earthquakes in this definite and positive sense which
alone deserves the name of forecasting? Unfortunately, no! All
reputable seismologists agree that we have no means at the
present time of arriving at a reliable forecast of any earthquake
anywhere.
The problem of earthquake forecasting has been under
intensive investigation in California and elsewhere for some
forty years; and we seem to be no nearer a solution of the
problem than we were in the beginning. In fact, the outlook is
much less hopeful.
In the California earthquake of 1906 the earth was torn by
a crack or fault which extended more than 150 miles in a
north-northwest south-southeast direction. All structures,
roads, fences, rows of trees, which crossed this fault were rent
apart and offset. The portion on the west side of the crack was
permanently shifted toward the north, relative to that on the
east side of the fault. The horizontal offset, or shift, between
the two parts originally continuous across the fault amounted
in one place to as much as twenty-one feet.
Immediately after the earthquake the United States Coast
and Geodetic Survey made a resurvey of the area and found
that points many miles west of the fault had been displaced
northward and points east of the fault had been displaced
42 1
422
J . B. Macelwutie
southward. These scientists also found that the regional shifts
were made up of two parts. There was a smaller shift which
had occurred sometime between the years 1866 and 1874, and
a larger shift that had taken place between 1892 and July,
1906. In the first of these intervals the earthquake of 1868 had
occurred, with relative movements on the Haywards fault east
of San Francisco Bay. In the second interval was the earthquake
of 1906, which involved the relative displacements we have
described along the San Andreas fault west of San Francisco
Bay.
To explain these distinct sets of movements, a well-known
seismologist proposed the theory of a slow northward creep
of the coastal region, which creep gradually distorted the rocks
in that portion of the earth’s crust until they were strained
beyond their capacity to resist. When this limit was reached,
fracture occurred along the weakest zone, and the rocks on
both sides of the fault rebounded elastically to new positions
of equilibrium. It was as if the Pacific coast started on a
vacation trip to Alaska and the rest of California refused to
go along and held back until the rock mantle tore.
“Ah!” said the seismologists, “here is a means of forecasting
earthquakes. Measure the gradual creep. When the shear
approaches the ultimate strength of the rocks, there will soon
be an earthquake.” So concrete pillars were placed on both
sides of the San Andreas fault and their relative positions were
determined very accurately with surveying instruments. The
relative positions have been redetermined again and again in
succeeding years, but no trace of creep has been observed.
“The pillars were placed too close together,” said the
seismologists. “Let us set up a precise network of triangulation
stations and lines of leveling and let us repeat the measurements
after a few years.” The United States Coast and Geodetic
Survey repeated its survey of California. This was done in the
early ’twenties. No sure sign of regional creep was found. Then
a very dense network of triangulation and leveling stations
was laid across the faults in southern California, and another
in central California. So far, no evidence of regional creep has
appeared.
The Japanese claimed to have observed a tilting of the
ground before an earthquake. American seismologists eagerly
seized on this as a possible means of forecasting earthquakes
and set up tiltmeters at Berkeley, California.
Tilts were observed, and so were earthquakes of very near
origin. But unfortunately there was no correlation between the
tilts and the earthquakes.
Do earthquakes come in cycles? Can they be triggered by
weather or tidal action? If so, they might be forecast. Many
statistical studies were made; but the conclusions were all
negative or very uncertain.
Every lead that was suggested has been followed and all
of them so far have led up blind alleys. Of course, earthquake prediction in a wider sense is possible. We know from
history and from the records of modern seismographs that
many more earthquakes occur in certain parts of the world
than in others. The rim of the Pacific Ocean is particularly
active. No year passes without a strong earthquake somewhere
in the lands that border the Pacific Ocean. The same is true
of most of the larger groups of islands in the Pacific. Hence
one needs only to make his prediction broad enough in space
and time and it is sure to be verified. But this is not earthquake
forecasting as we have defined it and as forecasting is usually
understood.
0 1997 Seismological Society of America, G J I 131, 421-422