The present volume differs from a text-book of seismology in giving brief, though detailed, accounts of individual earthquakes rather than a discussion of the phenomena and distribution of earthquakes in general. At the close of his Les Tremblements de Terre, Professor Fouqué has devoted a few chapters to some of the principal earthquakes between 1854 and 1887; and there are also the well-known chapters in Lyell's Principles of Geology dealing with earthquakes of a still earlier date. With these exceptions, there is no other work covering the same ground; and he who wishes to study any particular earthquake can only do so by reading long reports or series of papers written perhaps in several different languages. The object of this volume is to save him this trouble, and to present to him the facts that seem most worthy of his attention.
The chapter on the Japanese earthquake is reprinted, with a few slight additions, from a paper published in the Geographical Journal, and I am indebted to the editor, not only for the necessary permission, but also for his courtesy in furnishing me with clichés of the blocks which illustrated the original paper. The editor of Knowledge has also allowed me to use a paper which appeared four years ago as the foundation of the ninth chapter in this book.
CHARLES DAVISON.
Birmingham,
January, 1905.
I propose in this book to describe a few of the more important earthquakes that have occurred during the last half century. In judging of importance, the standard which I have adopted is not that of intensity only, but rather of the scientific value of the results that have been achieved by the study of the shocks. Even with this reservation, the number of earthquakes that might be included is considerable; and I have therefore selected those which seem to illustrate best the different methods of investigation employed by seismologists, or which are of special interest owing to the unusual character of their phenomena or to the light cast by them on the nature and origin of earthquakes in general.
Thus, the Neapolitan earthquake possesses interest from a historical point of view; it is the first earthquake in the study of which modern scientific methods were employed. The Ischian earthquakes are described as examples of those connected with volcanic action; the Andalusian earthquake is chiefly remarkable for the recognition of the unfelt earth-waves; that of Charleston for the detection of the double epicentre and the calculation of the velocity with which the vibrations travelled. In the Riviera earthquake are combined the principal features of the last two shocks with several phenomena of miscellaneous interest, especially those connected with its submarine foci. The Japanese earthquake is distinguished from others by its extraordinary fault-scarp and the very numerous shocks that followed it. The Hereford earthquake is a typical example of a twin earthquake, and provided many observations on the sound phenomena; while the Inverness earthquakes are important on account of their connection with the growth of a well-known fault. The great Indian earthquake owns few, if any, rivals within historical times, whether we consider the intensity of the disturbance or the diversity and interest of the phenomena displayed by it—the widespread changes in the earth's crust, both superficial and deep-seated, and the tracking of the unfelt pulsations completely round the globe.
Some terms are of such frequent use in describing earthquakes that it will be convenient to group them here for reference, others more rarely employed being introduced as they are required.
An earthquake is caused by a sudden displacement of the material which composes the earth's interior. The displacement gives rise to series of waves, which are propagated outwards in all directions, and which, when they reach the surface, produce the sensations known to us as those of an earthquake.
The region within which the displacement occurs is sometimes called the hypocentre, but more frequently the seismic focus or simply the focus. The portion of the earth's surface which is vertically above the seismic focus is called the epicentre. The focus and epicentre are often spoken of for convenience as if they were points, and they may then be regarded as the centres of the region and area in which the intensity was greatest. This is not quite accurate, but to attempt a more exact definition would at present be out of place.
An isoseismal line is a curve which passes through all points at which the intensity of the shock was the same. It is but rarely that the absolute intensity at any point of an isoseismal line can be ascertained, and only one example is given in this volume. As a rule, the intensity of a shock is determined by reference to the degrees of different arbitrary scales. These will be quoted when required.
Fig. 1.—Diagram to illustrate simple harmonic motion.ToList
In every strong earthquake there is a central district which differs in a marked manner from that outside in the far greater strength and complexity of the phenomena. As this district includes the epicentre, it is sometimes referred to as the epicentral area, but the term meizoseismal area is more appropriate, and will be employed accordingly.
The district over which an earthquake is perceptible to human beings without instrumental aid is its disturbed area. In like manner, that over which the earthquake-sound is heard is the sound-area.
A great earthquake never occurs alone. It is merely the most prominent member of a group of shocks of greater or less intensity, and is known as the principal shock or earthquake, while the others are called minor or accessory shocks, and fore-shocks or after-shocks according as they occur before or after the principal earthquake. When the sound only is heard, without an accompanying tremor being anywhere perceptible, it is more accurately called an earth-sound, but is frequently for convenience numbered among the minor shocks.
The movement of the ground during a vibration of the simplest character (known as simple harmonic motion) is represented in Fig. 1. The pointer of the recording seismograph is here supposed to oscillate along a line at right angles to AB, and the smoked paper or glass on which the record is made to travel to the left. The distance MP of the crest P of any wave from the line AB represents the amplitude of the vibration, the sum of the distances MP and NQ its range, and the length AB the period of the vibration. From the amplitude and period we can calculate, in the case of simple harmonic motion, both the maximum velocity and maximum acceleration of the vibrating particles of the ground.[1]
A few terms describing the nature of the shock are also in common use among Italians and Spaniards. An undulatory shock consists of one or several waves, the movement to and fro being along a nearly horizontal line; a subsultory shock of movements in a nearly vertical direction; while a vorticose shock consists of undulatory or subsultory movements crossing one another in different directions.
Earthquakes are grouped, according to their origin, into three classes. The first consists of slight local shocks, caused by the fall of rock in underground passages; the second of volcanic earthquakes, also local in character, but often of considerable intensity near the centre of the disturbed area; while in the third class we have tectonic earthquakes, or those directly connected with the shaping of the earth's crust, which vary in strength from the weakest perceptible tremor to the most destructive and widely felt shock. Of the earthquakes described in this volume, the Ischian earthquakes belong to the second class, and all the others to the third.
That tectonic earthquakes are closely connected with the formation of faults seems now established beyond doubt. They occur far from all traces of recent volcanic action. Their isoseismal lines are elongated in directions parallel to known faults, and this is sometimes the case in one and the same district with faults that occur at right angles to one another. Indeed, when several isoseismals are carefully drawn, it is possible from their form and relative position to predict the position of the originating fault.[2] The initial formation and further spreading of the rent may be the cause of a few earthquakes, but by far the larger number are due to the subsequent growth of the fault. The relative displacement of the rocks adjoining the fault, which may amount to thousands of feet, occasionally even to miles, is the result, not of one great movement, but of innumerable slips taking place in different parts of the fault and spread over vast ages of time. With every fault-slip, intense friction is suddenly brought into action by the rubbing of one mass of rock against the other; and, according to the modern view, it is this friction that gives rise to the earthquake waves.
In most earthquakes, the slip takes place at a considerable depth, perhaps not less than one or several miles, and the vertical slip is so small that it dies out before reaching the surface. But, in a few violent earthquakes, such as the Japanese and Indian earthquakes described in this volume, the slip is continued up to the surface and is left visible there as a small cliff or fault-scarp. In these cases, the sudden spring of the crust may increase and complicate the effects of the vibratory shock.
[1] If a is the amplitude of the vibration and T its period, the maximum velocity is 2πa ÷ T and the maximum acceleration 4π²a ÷ T².
[2] See Chapter VIII., on the Hereford and Inverness earthquakes.
Half a century ago, seismology was in its infancy. On the Continent, Alexis Perrey of Dijon was compiling his earthquake catalogues with unfailing enthusiasm and industry. In 1846, Robert Mallet applied the laws of wave-motion in solids, as they were then known, to the phenomena of earthquakes; and his memoir on the Dynamics of Earthquakes[3] may be regarded as the foundation-stone of the new science. During the next twelve years he contributed his well-known Reports to the British Association,[4] and prepared a series of instructions for the observation and study of earthquake-shocks.[5] The latter, it is worth noting, contains an outline, but hardly more than an outline, of the methods of investigation which he developed and employed eight years afterwards in studying the Neapolitan earthquake.
The history of Mallet's preparation for his great work is somewhat strange. No one else at that time possessed so full a knowledge of earthquake phenomena. It was, however, a knowledge that had little, if any, foundation in actual experience; for, when he was awakened by the British earthquake of November 9th, 1852, he failed to recognise its seismic character. Although this shock disturbed an area of about 75,000 square miles and was felt in all four parts of the kingdom, the paucity of observations and the absence of durable records combined in preventing the successful application of his new modes of study.[6] Nevertheless, with confidence unshaken in their power, he awaited the occurrence of a more violent shock, but five years had to pass before his opportunity came towards the close of 1857.
So destructive was the Neapolitan earthquake of this year (Mallet ranks it third among European earthquakes in extent and severity), that nearly a week elapsed before any news of it reached the outer world. Without further loss of time, he applied for and obtained a grant of money from the Council of the Royal Society, and proceeded early in the following February to what was then the kingdom of Naples. Armed with letters of authority to different officials, he visited the chief towns and villages in the meizoseismal area; and, in spite of unfavourable weather and the difficulties of travelling in a country so recently devastated, he completed his examination in little more than two months. It was a task, surely, that would have baffled any but the most enthusiastic investigator or one unspurred by the feeling that he possessed the key to one of the most obscure of Nature's problems.
Mallet's confidence in the accuracy of his methods was almost unbounded. His great report was published four years later; but he seems to have regarded it almost as a text-book of "observational seismology" and the results of his Neapolitan work as mere illustrations. His successors, however, have transposed the order of importance, and rank his two large volumes as the model, if not the inspirer, of many of our more recent earthquake monographs.
Fig. 2.—Isoseismal Lines of the Neapolitan Earthquake of 1857. (Mallet.)ToList
The position of the meizoseismal area, to which Mallet devoted most of his time, is indicated by the small oval area marked 1 in Fig. 2, represented on a larger scale in Fig. 9. It is 40 miles long and 23 miles wide,[7] and contains 950 square miles. Within this area, the loss of life was great and most of the towns were absolutely prostrated.
The next isoseismal, No. 2, which is also shown more clearly in Fig. 9, bounds the area in which the loss of life was still great and many persons were wounded, while large portions of the towns within it were thrown down. Its length is 65 miles, width 47 miles, and area 2,240 square miles. The third isoseismal includes a district in which buildings were only occasionally thrown down, though none escaped some slight damage, and in which practically no loss of life occurred. This curve is 103 miles long, 82 miles wide, and includes 6,615 square miles. Lastly, the fourth isoseismal marks the boundary of the disturbed area, which is 250 miles long, 210 miles wide, and contains not more than 39,200 square miles; an amount that must be regarded as strangely small, and hardly justifying Mallet's estimate of the Neapolitan earthquake as the third among European earthquakes in extent as well as in seventy.
As regards destruction to life and property, however, the Neapolitan earthquake owns but few European rivals. Less favourable conditions for withstanding a great shock are seldom, indeed, to be found than those possessed by the mediæval towns and villages of the meizoseismal area. In buildings of every class, the walls are very thick and consist as a rule of a coarse, short-bedded, ill-laid rubble masonry, without thorough bonding and connected by mortar of slender cohesion. The floors are made of planks coated with a layer of concrete from six to eight inches thick, the whole weighing from sixty to a hundred pounds per square foot. Only a little less heavy are the roofs, which are covered with thick tiles secured, except at the ridges, by their own weight alone. Thus, for the most part, the walls, floors, and roofs are extremely massive, while the connections of all to themselves and to each other are loose and imperfect.
Again, the towns, for greater security from attacks in early times, are generally perched upon the summits and steep flanks of hills, especially of the lower spurs that skirt the great mountain ranges; and the rocking of the hill-sites, in Mallet's opinion, greatly aggravated the natural effects of the shock. The streets, moreover, are steep and narrow, sometimes only five feet, and not often more than fifteen feet, in width; and the houses, when shaken down, fell against one another and upon those beneath them. As Dolomieu said of the great earthquake in 1783, "the ground was shaken down like ashes or sand laid upon a table."
Of the total amount of damage, not even the roughest estimate can be made. The official returns are clearly, and no doubt purposely, deficient, and obstacles were placed in Mallet's way when he endeavoured to ascertain the numbers of persons killed and wounded. Taking only the towns into account, he calculated that, out of a total population of 207,000, the number of persons killed was 9,589, and of wounded 1,343.[8] A few towns were marked by an excessively high death-rate. Thus, at Montemurro, 5000 out of 7002 persons were killed and 500 wounded; at Saponara, 2000 out of 4010 were killed; and, at Polla, more than 2000 out of a population of less than 7000.
The principal objects of Mallet's investigation were to determine the position of the epicentre and the depth of the seismic focus. If, in Fig. 3, F represents the seismic focus (here, for convenience, supposed to be a point), the vertical line FE will cut the surface of the earth in the epicentre E.[9] The dotted lines represent circles drawn on the surface of the earth with E as centre and passing through the places P and Q.
Fig. 3.—Diagram to illustrate wave-path and angle of emergence.ToList
When the impulse causing the earthquake takes place at the focus, two elastic waves spread outwards from it in all directions through the earth's crust. The first wave which reaches a point P consists of longitudinal vibrations, that is, the particle of rock at P moves in a closed curve with its longer axis in the direction FP. Mallet supposes this curve to be so elongated that it is practically a straight line coincident in direction with FP. In the second or transversal wave, the vibration of the particle at P takes place in a plane at right angles to FP. These vibrations Mallet, for his main purpose, neglects.
Returning to the longitudinal wave, Mallet calls the line FP the wave-path at P. The direction EP gives the azimuth of the wave-path, or its direction along the surface of the earth. The angle LPA, or EPF, he defines as the angle of emergence at the point P. If Q be farther from E than P, the angle EQF is less than the angle EPF, or the angle of emergence diminishes as the distance from the epicentre increases. At the epicentre, the angle of emergence is a right-angle; at a great distance from the epicentre, it is nearly zero.
Mallet argued that the direction of the wave-path FPA, or its equivalents, the horizontal direction EPL and the angle of emergence EPF, should be discoverable from the effects of the shock at P. The cracks in damaged buildings, he urged, would be at right angles to the wave-path FPA; overturned monuments or gate-pillars should fall along the line EPL, either towards or from the epicentre according to their conditions of support; loose or slightly attached bodies, such as the stone balls surmounting gate-pillars, should be projected nearly in the direction of the wave-path FPA, and their subsequent positions, supposing the balls not to have rolled, should give the horizontal direction EPL of the wave-path, and might, in some circumstances, determine the angle of emergence and the velocity with which they were projected. I shall return to details later on. For the present, it is clear that, in the destruction wrought by the earthquake, Mallet expected to find the materials most valuable for his purpose. Indeed, so obvious did this mode of examination appear to him, that he could not conceal his surprise at the blindness of his predecessors. They seem, he says, "to have been perfectly unconscious that in the fractured walls and overthrown objects scattered in all directions beneath their eyes, they had the most precious data for determining the velocities and directions of the shocks that produced them."
Mallet's Method of Determining the Position of the Epicentre.—In many cases the examination of a damaged building or of an overthrown body served more than one purpose, providing materials for ascertaining the depth of the seismic focus as well as the position of the epicentre. For the present, however, it will be convenient to consider alone the method by which the latter object was to be attained.
Fig. 4.—Diagram to illustrate Mallet's method of determining position of epicentre.ToList
Nothing could be simpler than the principle of the method proposed. The horizontal direction PL of the wave-path at any place P (Fig. 4), when produced backwards, must pass through the epicentre E; and the intersection of the directions at two places, P and Q, must therefore give the position of the epicentre. In practice, it is of course impossible to determine the direction with very great accuracy, and Mallet therefore found it necessary to make several measurements in every place, and to visit all the more important towns within and near the meizoseismal area.
In a ruined town there are many objects from which the direction may be ascertained, the most important of all, according to Mallet, being fissures in walls that are fractured but not overthrown. He regarded such fissures, indeed, as "the sheet-anchor, as respects direction of wave-path, to the seismologist in the field," and at least three out of every four of his determinations of the direction were made by their means. If the buildings are detached and large, simple and symmetrical in form, well built and not too much injured, the fissures in the walls should, he argued, occur along lines at right angles to the wave-path, whether that path be parallel or inclined to the principal axis of the building. Cracks in the floors and ceilings should also be similarly directed, and provide evidence which Mallet regarded as only second in value to that given by the walls.
Fig. 5.—Plan of Cathedral Church at Potenza. (Mallet.)ToList
No building showed the different kinds of evidence on which Mallet relied as clearly as the cathedral church at Potenza, the plan of which is given in Fig. 5, and the vertical section along its axis in Fig. 12. This is a modern work, nearly 200 feet long, with its axis directed east and west. The walls are composed of fairly good rubble masonry and brick; and the arches in the nave and transepts, the semi-cylindrical roof and the central dome are made of brick. The fissures represented in both diagrams were drawn to scale by the cathedral architect before Mallet's arrival, and, as the work of an unbiassed observer, are of special value. Most of those in the roof, it will be seen, were transverse to the axial line of the church; but there were others parallel to this line, one in particular running right along the soffit of the nave and chancel. There were also numerous small fissures in the dome, due to local structural causes and therefore of varying direction, and a large portion of the dome slipped westward, leaving open fissures of seven to eight inches in width. The mean direction of the wave-path, as deduced from nine sets of fissures, none of which differs more than four degrees from the mean, is W. 2½° S. and E. 2½° N., which corresponds precisely with the direction of throw on the displaced portion of the dome. The great east and west fissures in the arch of the nave and chancel Mallet attributed to a second shock, of the existence of which there is ample evidence.
Fig. 6.—Fallen gate-pillars near Saponara. (Mallet.)ToList
Next to fissures, Mallet made most use of overthrown objects, such as the two gate piers near Saponara, represented in Fig. 6. They were made of rubble ashlar masonry, three feet square and seven feet in height. Both were fractured clean off at the level of the ground, the mortar being poor, and fell in directions that were accurately parallel, indicating a wave-path towards S. 39½°E. A few observations were also made on projected stones, fissures in nearly level ground, and the swinging of lamps and chandeliers; but their value was small, except as corroboration of the more important evidence afforded by fissures in the walls and roofs of buildings.
Remarks on Mallets Method.—It would have been more difficult in Mallet's day than it is now, to offer objections to his method of determining the position of the epicentre. The focus, as he was well aware, could not be a point, and, at places near the epicentre (the very places where most of his observations were made), there must be rapid changes of direction due to the arrival of vibrations from different parts of the focus. He records the occurrence of the so-called vorticose shocks at several places, though he attributes them to another cause. Perhaps the best known example of such a shock is that which has been so well illustrated by the late Professor Sekiya's model of the motion of an earth-particle during the Japanese earthquake of January 15th, 1887. The motion in this case was so complicated that the model was, for simplicity, made in three parts, the first of which alone is represented in Fig. 7.[10] It is clear that in such an earthquake, Mallet's method would utterly fail in giving definite results.
While this shock was one of great complexity, another Japanese earthquake, that of June 20th, 1894, was unusually simple in character. The movement at Tokio consisted of one very prominent oscillation with a total range of 73 mm. or 2.9 inches in the direction S. 70° W.; the vibrations which preceded and followed it being comparatively small. Most, if not all, of the damage caused by the earthquake must have been due to this great oscillation; and yet the cylindrical stone-lamps so common in Japanese gardens were found by Professor Omori to have fallen in many different directions. Taking only those which had circular bases, twenty-nine were overthrown in directions between north and east, sixteen between east and south, eighty-one between south and west, and fourteen between west and north.[11] Fig. 8 represents Professor Omori's results graphically, the line drawn from O to any point being proportional to the number of lamps which fell in directions between 7½° on either side of the line.
Fig. 7.—Model to illustrate the motion of an earth-particle during an earthquake. (Sekiya.)ToList
Fig. 8.—Plan of directions of fall of overturned stone-lamps at Tokio during the earthquake of 1894.ToList
It will be seen from this figure that most of the stone lamps fell in directions between west and south-west, and it is remarkable that the mean direction of fall is S. 70° W.,[12] which is exactly the same as that of the great oscillation. Somewhat similar results were obtained by this able seismologist at different places affected by the great Japanese earthquake of 1891 (Figs. 43 and 44), and the study of the apparent directions observed during the Hereford earthquake of 1896 leads to the same conclusion.
It thus appears that an isolated observation may give a result very different from the true direction. Indeed, if we may judge from Professor Omori's measurements in 1894, the chance that a single direction may be within five degrees of the mean direction is about 1 in 9. But, on the other hand, it is equally clear from these and other observations that the mean of a large number of measurements will give a result that agrees very closely with the true direction.
One other point may be alluded to before leaving Professor Omori's interesting observations. It would seem, from the list that he gives, that he exercised no selection in his measurements, but continued measuring the direction of every fallen lamp indifferently until he had obtained sufficient records for his purpose. Now, if the number of fallen lamps at his disposal had been small, say 12 instead of 144, the mean observed direction would probably have differed from the direction given from the seismograph.[13] But, on the other hand, a preliminary survey without any actual measurements would have revealed at once the predominant direction of overthrow, and a fairly accurate result might have been obtained by neglecting discordant directions and taking the mean of those only which appeared to agree with the mentally determined average.
This, indeed, appears to have been the course followed, more or less unconsciously, by Mallet in his Neapolitan work. "When the observer," he says, "first enters upon one of those earthquake-shaken towns, he finds himself in the midst of utter confusion. The eye is bewildered by 'a city become an heap.' He wanders over masses of dislocated stone and mortar, with timbers half buried, prostrate, or standing stark up against the light, and is appalled by spectacles of desolation.... Houses seem to have been precipitated to the ground in every direction of azimuth. There seems no governing law, nor any indication of a prevailing direction of overturning force. It is only by first gaining some commanding point, whence a general view over the whole field of ruin can be had, and observing its places of greatest and least destruction, and then by patient examination, compass in hand, of many details of overthrow, house by house and street by street, analysing each detail and comparing the results, as to the direction of force, that must have produced each particular fall, with those previously observed and compared, that we at length perceive, once for all, that this apparent confusion is but superficial."
Fig. 9.—Meizoseismal area of Neapolitan earthquake. (Mallet.)ToList
Mallet's Determination of the Epicentre.—Within the third isoseismal line Mallet made altogether 177 measurements of the direction of the wave-path at 78 places. These are plotted on his great map of the earthquake; but, owing to the small scale of Fig. 9, it is only possible to represent, by means of short lines, the mean or most trustworthy direction at each place.[14] Producing these directions backwards, he found that those at sixteen places passed within five hundred yards of a point which is practically coincident with the village of Caggiano; those at sixteen other places passed within one geographical mile (1.153 statute miles) of this point; the directions at sixteen more places within two and a half geographical miles; while those at twelve places passed through points not more than five geographical miles from Caggiano. As the direction of the shock at places near the epicentre must have been influenced by the mere size of the focus, this approximate coincidence is certainly remarkable, and there can be little doubt, I think, that the epicentre, or, at any rate, an epicentre must have been situated not far from the position assigned to it by Mallet's laborious observations.
Existence of Two Epicentres.—It is difficult, however, to realise that the impulse at the focus corresponding to Mallet's epicentre was the origin of all the destruction of life and property that occurred. The position of the epicentre close to the north-west boundary of the meizoseismal area, the extraordinary extension of that area towards the south-east, and especially the great loss of life at Montemurro and the adjoining towns, can hardly be accounted for in this manner. Mallet himself recognised that these facts required explanation, and he suggested that the situation and character of the towns were in part responsible for their ruin, and the physical structure of the country for the course of the isoseismal lines. But the comparative escape of places much nearer Caggiano, and the wide extent of the meizoseismal area, embracing many towns and villages of varied character and site and many different surface-features, point unmistakably to a different explanation.
Fig. 10.—Distribution of death-rate within meizoseismal area of Neapolitan earthquake.ToList
One clue to the solution of the problem is afforded by the seismic death-rate of the damaged towns. From a table given by Mallet (vol. ii. pp. 162-163), we know the population before the earthquake of the different communes in the province of Basilicata, and the loss of life in each due to the shock; and from these figures we can find the percentage of deaths at nearly every place of importance. As will be seen from Fig. 10, it varies from seventy-one at Montemurro and fifty at Saponara down to less than one at all the places marked to which figures are not attached. There is thus a group of places, with its centre near Montemurro, where the loss of life far exceeded that in the surrounding country; and also a slightly less-marked group, with its centre near Polla, in the north-west of the meizoseismal area; while in the intermediate region the death-rate was invariably small. Too much stress should not be laid upon the exact figures, for there were no doubt local conditions that affected the death-roll. But it seems clear that one focus was situated not far from Montemurro; while the north-westerly group of places, combined with Mallet's observations on the direction, point to a second focus near Polla, about twenty-four miles to the north-west. It will be seen in a later section that the observations on the nature of the shock also imply the existence of a double focus.
Mallet's Method of Determining the Depth of the Focus.—In ascertaining the position of the epicentre, Mallet's work was remarkable only for the novelty of the method employed by him; but, in his attempt to calculate the depth of the seismic focus, he was breaking new ground. That the depth must be comparatively small had already been recognised, and was indeed obvious from the limited area disturbed by nearly every earthquake. No one, however, had tried to estimate the depth in miles; and it is impossible not to sympathise with Mallet while he accumulated his observations with feverish activity and subjected them to the first rough examination even if one cannot share his confidence that he had succeeded in measuring the depth "in miles and yards with the certainty that belongs to an ordinary geodetic operation."
Fig. 11.—Diagram to illustrate Mallet's method of determining depth of seismic focus.ToList
The method employed by him for the purpose is no less simple theoretically than that used for locating the epicentre. If the position of the latter (E) is known, one accurate measurement of the angle of emergence EPF, at any other point P would be sufficient to fix the depth of some point within the focus F (Fig. 11). Here, again, Mallet relied chiefly on fissures in walls that were fractured but not overthrown. In detail, these fissures are nearly always jagged or serrated, for they tend to follow the lines of joints rather than break through the solid stone, though they sometimes traverse bricks and mortar alike. But the general course of the fissures, he urged, would be at right angles to the wave-path, and their inclination to the vertical should be equal to the angle of emergence.
In obtaining measurements of this angle, the buildings to be chosen are those of large size, with few windows or other apertures, and with walls made of brick or small short-bedded stones. The cathedral-church at Potenza perhaps satisfies these conditions more closely than any other structure examined by Mallet. The plan of the fissures in the walls and roof has been given in Fig. 5, and Fig. 12 represents the fissures In the vertical section along the axial line and looking north, as drawn by the cathedral architect. From these fissures Mallet calculated the mean angle of emergence at Potenza to be 23° 7'. The distance of Potenza from Caggiano being seventeen miles, and the height of the former being 2,580 feet, the depth of the focus resulting from this observation alone would be 6¾ miles below the level of the sea.
Fig. 12.—Vertical section of Cathedral Church at Potenza. (Mallet.)ToList
Objection to Mallet's Method.—The weakest point in Mallet's method is probably his assumption that the wave-paths are straight lines extending outward from the focus. Even if the depth of the focus is not more than a few miles, the waves must traverse rocks of varying density and elasticity, and, at every bounding surface, they must undergo refraction. If the rocks are so constituted that the velocity of the earth-waves in them increases with the depth, then the wave-paths must be bent continually outwards from the vertical, so that the angle of emergence at the surface may be considerably less than it would have been with a constant velocity throughout. In this case, the actual depth will be greater, perhaps much greater, than the calculated depth. For instance, if the angle of emergence at Potenza were diminished only 5° by refraction, the calculated depth of the focus would be too small by 1¾ miles.
Fig. 13.—Diagram of wave-paths at seismic vertical of Neopolitan earthquake. (Mallet.)ToList
Mallet's Estimate of the Depth of the Focus.—Mallet measured the angle of emergence at twenty-six places, the mean angle (i.e. the mean of the greatest and least observed angles) varying from 72° at Vietri di Potenza and 70° degrees at Pertosa, which are about two miles from the calculated epicentre, to 11½° at Salerno, distant about 40 miles. Fig. 13 reproduces part of the diagram on which he plotted the mean angle of emergence at different places. The horizontal line represents the level of the sea, and the vertical line one passing through the epicentre and focus, called by Mallet the "seismic vertical." The lines on the left-hand side represent the commencing wave-paths (assumed straight) to the observing stations situated to the westward of the meridian through the epicentre, those on the right-hand side corresponding to places to the eastward of the same meridian. Small horizontal marks are added to indicate the depth in miles below the level of the sea.
It will be seen, from this diagram, that all the wave-paths start from the seismic vertical at depths between three and nine miles; but the points of departure are clustered thickly within a portion, the length of which is about 3½ miles and the mean depth about 6½ miles. So great was Mallet's confidence in these calculations that he assigns the diverging origin of the wave-paths to different points of the focus, and thus concludes that, while the mean depth of the focus was about 6½ miles, its dimensions in a vertical direction did not exceed 3½ miles.
How far Mallet's results should be accepted as correct, it is difficult to say in our ignorance of the constitution of the earth's interior. There can be no doubt that the focus was of considerable size, and that, in consequence, the wave-paths would diverge from different points of it. But that each wave-path should actually intersect the focus, and so enable its magnitude to be determined, would surely involve an approach to some law connecting the direction of a wave-path with the depth of its own origin, and no such law seems to be ascertainable. Nor can the limitation of these apparent origins between certain depths be held to argue that the focus, or any part of it, was equally confined, for the wave-paths would to a great extent be similarly refracted. I fear that the only conclusions that we can with safety draw from Mallet's admirable work are that his figures indicate the order of magnitude both of the vertical dimensions and of the mean depth of the focus.
It is not easy to form any precise image of the earthquake as it appeared to the terrified witnesses within the meizoseismal area. To minds unbalanced by the suddenness of the shock and by the crash of falling houses, actuated too by the intense need of safety, the mere succession of events must have presented but little interest. The interval of two months that elapsed between the occurrence of the earthquake and its investigation was also unfavourable to the collection of accurate accounts from a wonder-loving people. Only one feature, therefore, stands out clearly in the few records given by Mallet—namely, the division of the shock into two distinct parts.
In the central district, this division is perhaps less apparent than elsewhere. At Polla, for instance, which lies close to the north-west epicentre, the first warning was given by a rushing sound; almost instantly, and while it was yet heard, came a strong subsultory or up-and-down movement, succeeded after a few seconds, but without any interval, by an undulatory motion. At Potenza, which is not far from the same epicentre but a few miles outside the meizoseismal area, the separation was more pronounced. According to one observer, the first movement was from west to east; and, within a second or two afterwards, there was a less violent shock in a transverse direction, followed immediately by a shaking in all directions, called by the Italians vorticose. Naples lies sixty-nine miles from the north-west epicentre, and here more accurate observations could be made. Dr. Lardner, well known fifty years ago as a writer of scientific works, describes the first movement felt there as "a short, jarring, horizontal oscillation, that made all doors and windows rattle, and the floors and furniture creak. This ceased, and after an interval that seemed but a few seconds was renewed with greater violence, and, he thought, with a distinctly undulatory movement, 'like that in the cabin of a small vessel in a very short chopping sea.'"
In five other earthquakes studied in this volume, the separation of the shock into two parts was a well-marked phenomenon. In the Neapolitan earthquake, the separation was so distinct that Mallet took some pains to account for its origin. He regarded it in every case as due to the reflection or refraction of the earth-waves by underlying rocks, though he does not explain why the reflected or refracted wave should be more intense than that transmitted directly. I shall refer to the subject in greater detail when describing the Andalusian, Charleston, Riviera, and Hereford earthquakes. For the present, it may be sufficient to urge that the double shock cannot have been due to the separation of the original waves by underground reflection or refraction, for then the second part should have been generally the weaker; nor to the succession of longitudinal and transverse waves, for, in that case, every earthquake-shock should be duplicated. The only remaining supposition is that there was a second impulse occurring either in the same or in a different focus.
Which alternative should be adopted, the evidence on the nature of the shock is too scanty to determine. The defect is, however, supplemented by Mallet's observations on the direction of motion; for, at many places within and near the meizoseismal area, he met with the clearest signs of a double direction. Sometimes this was apparent to the senses of the observer; in other cases, damaged buildings presented two sets of fissures. At La Sala and near Padula, the first movement was roughly east and west, the second north and south. At Moliterno, there was evidence of a subordinate shock at right angles to the chief one; in the neighbourhood of Tramutola, its direction was from about E. 30° S. In these and other cases, Mallet saw the effects of earthquake-echoes; but the underground reflection of earth-waves would give rise to the second part of the shock, not the first as at La Sala and Padula. Moreover, the secondary directions, though they are seldom recorded accurately, point nearly to an epicentre not far from Montemurro. The observations on the nature and direction of the double shock thus confirm the conclusion, derived from the distribution of the seismic death-rate, that there were two detached foci, one near Polla and the other near Montemurro.
This seems to be the best explanation of the facts recorded by Mallet. There is, however, a possible difficulty that should not be overlooked—namely, the apparently slight influence of the Montemurro focus on the mean direction of the shock (Fig. 9). At a few places, of course, the mean direction passes through both epicentres; at some others, as we have seen, one of the two observed directions points towards the Montemurro epicentre. It is not impossible, also, that Mallet, after the first few days' work, may occasionally have quite unconsciously selected and measured those fissures from the maze presented to him which agreed most closely with his early impressions obtained from the neighbourhood of Polla. But, for places nearer Polla than Montemurro (and these form the majority of those visited by Mallet), the probable explanation of the difficulty is that the Montemurro focus was not so deep as the Polla focus. This, as will appear more fully in the next chapter, would account for the comparatively great intensity in the immediate neighbourhood of Montemurro and for its rapid decline outwards; and it receives some support from an isolated reference by Mallet to two angles of emergence at Padula, one of 25° from the north, and the other of 8° or 10° in the perpendicular walls.
The elements of the wave-motion, as mentioned in the introductory chapter, are four in number, namely, the period, amplitude, maximum velocity, and maximum acceleration. If any two of these are known for each vibration—and the first two are now given by every accurately constructed seismograph—the others can be determined if the vibrations follow the law of simple harmonic motion.[15]
Amplitude.—To ascertain the amplitude, Mallet had to rely chiefly on the fissures made in very inelastic walls. If the parts into which such a wall are fractured are free to move, and yet, being inelastic, obliged to remain in the farthest position to which they are carried by the wave, the distance traversed by the centre of gravity of one of the displaced parts should give a "rude approximate measure" of the horizontal amplitude of the earth-wave. At Certosa, near Padula, he thus found the amplitude to be about 4 inches, at Sarconi about 4¾ inches, and at Tramutola about 4½ inches. From somewhat similar evidence, the amplitude at Polla appears to have been about 2½ or 3 inches; and, from the oscillation of a suspended clock or watch on a rough wall, about 3½ inches at La Sala and 1¾ inches at Barielle. With the exception of Barielle, these places lie nearly on a straight line passing through Mallet's epicentre, and he gives the following table, showing an increase in amplitude with the distance from the epicentre:—
| Polla. | La Sala. | Certosa. | Tramutola. | Sarconi. | |
| Distance in miles | 4.0 | 13.4 | 19.0 | 23.8 | 30.8 |
| Amplitude in inches | 2½ | 3½ | 4 | 4½ | 4¾ |
The existence of the Montemurro focus must, however, complicate any relation that may connect these two quantities.
Maximum Velocity.—The means at Mallet's disposal for determining the maximum velocity were more numerous than those available for the amplitude. From the dimensions of a fallen column of regular form we should be able, he remarks, to find an inferior limit to the value of the maximum velocity; while a superior limit at the same place may be obtained from some other regular solid which escaped being overthrown. If a loose body is projected by the shock at a place where the angle of emergence is known, the horizontal and vertical distances traversed by the centre of gravity will give the velocity of projection. Or, if two such bodies are projected at one place, the same measures for each will as a rule give both the angle of emergence and the velocity of projection. A third method depends on the fissuring of walls, supposing that we know the force per unit surface which, when suddenly applied, is just sufficient to produce fracture. Sometimes more than one method must be applied to the same object. The two gate-pillars near Saponara (illustrated in Fig. 6) for example required a horizontal velocity of 5.48 feet per second to fracture them, and an additional velocity of 5.14 feet per second to overthrow them.
The well-known seismologist, Professor Milne, urges very forcibly that measurements obtained from the projection or fall of columns are unreliable, for the earlier tremors might cause the columns to rock, and their overthrow need not therefore measure accurately the maximum velocity of the critical vibration.[16]