Cambridge fcologicai Series A Manual of Seismology 'H>S'^r.'y.':fvMii^^Seciaiiiy-i><-Sf^^^^ THE LIBRARY OF THE UNIVERSITY OF CALIFORNIA LOS ANGELES GIFT OF R. L. Troxell CAMBRIDGE GEOLOGICAL SERIES A MANUAL OF SEISMOLOGY CAMBRIDGE UNIVERSITY PRESS C. F. CLAY, Manager LONDON : FETTER LANE, E.C. 4 LONDON : H. K. LEWIS ANl) 00., Ltd., 136, Gower Street, W.C. 1 NEW YORK : THE MACMILLAN CO. BOMBAY 1 CALCUTTA \ JL\CMILLAN AND CO., Ltd. MADRAS J TORONTO : THE MACMILLAN CO. OF CANADA, Ltd. TOKYO : MARUZEN-KABUSHIKLKAISHA ALL BIGHTS RESERVED A MANUAL OF SEISMOLOGY BY CHARLES DAVISON, Sc.D. CAMBRIDGE AT THE UNIVERSITY PRESS 1921 MlKTItU CM eRCAT B^lrTAiH. PREFACE MY chief difficulty in writing this textbook has been to decide on what should be included and what excluded, to determine in fact the relative importance of different branches of a subject wliich is of growth so modern that its treatment has not yet become defined. Until near the close of the last century, seismology was regarded as a department of geology, but, while its growth in that direction has by no means ceased, the more recent advances have been largely the work of mathe- maticians and physicists. As this volume belongs to a series of geological manuals, much of that work is here left unnoticed, and readers who wish to knoAv more of these recent de^^elop- ments may be referred to Prof. Knott's Physics of Earthquake Phenomena (Oxford University Press) and Dr G. W. Walker's Modern Seismology (Longmans). Partly in order to save room, I have omitted several subjects which are usually included within the domain of seismology. vSome of these (such as the bending of the earth's crust by tidal loading and the occurrence of man}' micro-tremors) have in reality no connexion with earthquakes. Others (such as the effects of earthquakes on men and animals, the rotation of columns and experiments on the velocity of earth-waves) are of interest but are not essential to the subject. Though a few historical notes are inserted here and there, no attemjjt is made to provide a history of seismology. My aim Jias been to give an outline of our present knowledge, which of course is usually in advance of the work of pioneers. The number of references to any writer in the index is thus rather a testimony to the recency of his achievements than to the actual part which he has taken in promoting the science. Otherwise, there would have been many more references to the work of Perrey. Mallet and Milne. Readers who desire further information on the history of the science may consult a series of pajiers on the "Founders of Seismology" in the Geological Magazine for 1921. aS vi PREFACE With regard to references generally, it may be convenient if I mention here the plan which I have followed. At the beginning of nearly every chapter, I have given a list of the memoirs which the reader who desires fuller information may study with advantage. Throughout the chapter, these are quoted under the author's name. References required in each section are collected in a footnote at the end of the section, and, when- ever a choice is available, I have preferred those to English or easily accessible works. The abbreviated titles of the more im- portant books and journals are given after the Table of Contents. For two of the diagrams (Figs. 11 and 21), I am indebted to the courtesy of the Directors of the Cambridge Scientific Instru- ment Company. CHARLES DAVISON. Cambridge. March, 1921. CONTENTS CHAP. I. INTRODUCTION PAGE 1 II. SEISMOGRAPHS 6 The Essential Parts of a Seismograph .... 8 Seismographs for Near Earthquakes . . . . 17 Seismographs for Distant Earthquakes . . . . 22 III. NATURE AND INTENSITY OF EARTHQUAKE- MOTION 30 Nature of Earthquake-Motion ..... 30 Elements of Earthquake-Motion ..... 36 Intensity of Earthquake-Motion ..... 38 Scales of Seismic Intensity ...... 44 Isoseismal Lines and Disturbed Area .... 48 Direction of Earthquake-Motion ..... 50 Duration of Earthquake-Motion ..... 53 IV. SOUND-PHENOMENA OF EARTHQUAKES . . 56 V. DEFORMATIONS OF THE EARTH'S CRUST . . 69 Fault-Displacements ....... 70 Nature of Fault-Deformation . . . . . 81 Warping 87 VI. SEISMIC SEA-WAVES . . . . . . . 90 VII. SECONDARY EFFECTS OF EARTHQUAKES . 101 Landslips ......... 101 Compression of Alluviuni ...... 104 Earth-Fissures . . . . . . . .10.5 Effects on Underground-Water ..... 109 Sand-Vents Ill Rise of River-Channels, etc. . . . .113 Effects of Earthquakes on Glaciers . . . . 114 VIII. POSITION OF THE SEISMIC FOCUS . . . . 110 Position of the Epicentre . . . . . .110 Depth of the Seismic Focus . .124 Vlll CONTENTS CHAP. IX. XI. XII. PROPAGATION OF EARTHQUAKE-WAVES Reflection and Refraction of Earthquake- Waves Earthquake-Motion at Great Distances from the Origin Nature of the Earth's Interior .... Determination of the Epicentre of a Distant Earthquake GEOGRAPHICAL DISTRIBUTION OF EARTHQUAKES Construction of Seismic Maps Seismic Maps of the World . I^aws of Seismic Distribution Migrations of Seismic Activity Distribution of Submarine Earthquakes FREQUENCY AND PERIODICITY OF EARTHQUAKES Frequency of Earthquakes . Periodicity of Earthquakes . ACCESSORY SHOCKS . Fore-Shocks . . . After-Shocks .... Sympathetic Shocks XIII. VOLCANIC EARTHQUAKES .... Relations between Tectonic and Volcanic Earthquakes Earthquakes of Active Volcanoes .... Earthquakes of Dormant and Extinct Volcanoes Characteristics of Volcanic Earthquakes Origin of Volcanic Earthquakes .... XIV. ORIGIN OF TECTONIC EARTHQUAKES . Earthquakes and the Growth of Faults Origin of Simple Earthquakes Origin of Twin Earthquakes . Origin of Complex Earthquakes Origin of Accessory Shocks . Origin of Earthquake-Sounds Conclusion .... INDEX 251 LIST OF ILLUSTRATIONS FIG. 1 2-4 5-7 8 9 10 11 12 1:3 14 15 lU IT 1« IJJ 20 21 22 23 24 25 2G 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 PAGE Diagram of simple harmonic vibration .... 2 Diagrams illustrating the theory of the common, inverted and horizontal pendulums ....... 9 Supports of horizontal pendulums . . . . . 13 Record of the Derby earthquake of 1903 obtained with an un- damped pendulum . . . . . . . 16 Evving horizontal motion seismogra|>h . . . . 17 Ewing Vertical inotion seismograph ..... 18 Ewing three-component seismograph ..... 19 Ewing duplex-pendulum seismograph . . . . . 21 Darwin bifilar pendulum ....... 22 Milne seismograph . . . . . . . .23 Omori horizontal pendulum ...... 26 Marvin inverted pendulum ...... 28 Vibrations of simple and twin earthquakes .... 31 Record of .Japanese earthquake of .Jan. 15, 1887 (Sekiya) . 33 liecord of Tokyo earthquake of .June 20, 1894 (Omori) . . 34 Record of .Japanese earthquake of .July 27, 1905 (Omori) . 35 Record of a .Japanese earthquake on a stationary plate . 36 Overturning of columns ....... 39 Construction of isoseismal lines ...... 47 Isoseismal lines of the Charleston earthcjuake of 188(5 (l)utton) 48 Direction-rose for the Tokyo earthquake of .June 20, 1894 . 51 Map of the directions of the shock and of the absolute iso- seismal lines of the Mino-Owari earthqxiakc of 1891 (Omori) ,52 Relation between the duration of the preliminary tremor and the distance of the station from the origin (Omori) . . 54 Map of the Derby earthquake of March 24, 1903 ... 60 Map of the Bolton earthquake of Feb. 10. 1889 ... 66 Map of the Ilclston earthquake of April 1, 1898 ... 67 Fault-scarp of the Mino-Owari eartlujuakc of 1891 (Koto) . 71 Secondary cracks, Californian carth(|uake of 190() (Omori) . 72 Minor fault, Alaskan cartlu|uake of 1899 (Tarr an (Omori) . 86 Diagram illustratinjr the displacements in the Formosa earth- quake of l!t()(J (Omori) 87 Sanriku carth(piakc (189(i) sea-waves recorded at .\yukawa (Honda) \ . 94 I(|ui(|ue eartliquake (1877) sea-waves recorded at .San Fran- cisco (Milne). . ... . . . . . 95 Ilill-foot fissures, Assam earthquake of 1897 (Oldham) . 108 X LIST OF ILLUSTRATIONS FIG. PAGE 45 Fault-block fissures, New Madrid earthquake of 1812 (Fuller) 109 46 Sand- vent, Assam earthquake of 1897 (Oldham) . . .112 47 Epieentral area of the Assam earthquake of 1897 (Oldham) . 117 48 Omori's method of determining the epicentre . . . 118 49 Mallet's method of determining the epicentre of the Neapolitan earthquake of 1857 120 50 Map of the Inverness earthquake of 1901 . . . . 121 51 Map of the Derby earthquake of 1904 .... 122 52 Map of the Kangra earthquake of 1905 (Middlemiss) . . 123 53 Von Seebach's time-curve . . . . . . .125 54 Diagram illustrating Mallet's and Dutton's methods of deter- mining the depth of the focus from variations of intensity . 130 55 Diagram illustrating the relative depths of the foci of two earthquakes determined by the more or less rapid variations of intensity . . . . . . . . .132 56 Diagram representing the energies of the reflected and re- fracted waves, rock to water, incident wave condensational (Knott) 137 57 Diagram representing the energies of the reflected and re- fracted waves, rock to water, incident wave distortional (Knott) 137 58 Diagram representing the directions and energies of the re- flected and refracted waves, rock to water, incident wave condensational . . . . . . . .138 59 Diagram representing the directions and energies of the re- flected and refracted waves, rock to water, incident wave distortional . . . . . . . . .139 60 Seismogram of the Californian earthquake of April 18, 1906, at Edinburgh (Milne seismograph) . . . . .142 61 Seismogram of the Kangra earthquake of April 4, 1905, at Birmingham (Omori horizontal pendulum) . opposite 142 62 Seismogram of the Asia Minor earthquake of Feb. 9, 1909, at Pulkowa (Galitzin seismograph) . . . oppositk 142 68 Seismogram of the Bonin Islands earthquake of Nov. 24, 1914, at Bidston (Milne-Shaw seismograph) . . opposite 142 64 Time-curves of the P, S and L waves . . . .146 65 Forms of seismic rays (Knott) . . . . . .151 66 Diagram representing the variation of the primary and second- ary wave-velocities with the depth (Knott) . . .152 67 Diagram illustrating the internal reflection of earthquake- waves at the earth's surface . . . . . .153 68 Diagram illustrating the returns of the long waves . . 155 69 Diagram illustrating the determination of the epicentre from the times of arrival of the long waves . . . . 158 70 Seismic map of Japan (1885-1892) 164 71 Milne's seismic map of the world . . . . .166 72 Map of seismic regions in .lapan (Omori) .... 170 73 Distribution of earthquake-zones in Calabria (Baratta) . . 172 74 Distribution of earthquakes in central .lapan (Omori) . . 173 75 Principal seismic regions of the equatorial Atlantic (Rudolph) 175 76 Monthly variation of seismic frequency in (i) Austria and (ii) Hungary, Croatia and Transylvania . . . .184 77 Diurnal variation of seismic frequency in (i) Switzerland and (ii) the West Indies 184 LIST OF ILLUSTRATIONS xi FIG. PAGE 78 Harmonic curves and their compound . . . .185 79 Annual periodicity of earthquakes in (i) Austria, (ii) Hungary, Croatia and Transylvania, and (iii) Switzerland and the Tyrol .• 189 80 Variation of annual seismic maxinium-epochs in Japan . . 192 81 Diurnal periodicity of the after-shocks of the Mino-Owari earthquake of Oct. 28, 1891, at Nagoya, (i) Oct. 29-Nov. 10, 1891, (ii) Nov. 11, 1891-Dec. 31, 1899 . . . . 19G 82 Course of the fault-scarj), etc., of the Mino-Owari earthquake of 1891 201 83 Distribution of the fore-shocks of the Mino-Owari earthquake of 1891 203 84 Decline in frequency of the after-shocks of the Mino-Owari earthquake of 1891 207 8.5 Percentage of feeble after-shocks of the Mino-Owari earth- quake of 1891 at Gifu 209 80 Distribution of the after-shocks of the Mino-Owari earthquake of 1891 (Nov .-Dec. 1891) 209 87 Distribution of the after-shocks of the Mino-Owari earthquake of 1891 (.Tuly-Aug. 1892) 210 88 \olcanic chain of south .Japan (Omori) . . . . 2 1 fi 89 Map of the Fondo Macchia (Etna) earthquakes of 1865 and 1911 (Baratta and Ricc6) 218 90 Map of the Lincra (Etna) earthquakes of May 8, 1914 (Pla- tania) ' . . . 219 91 .Seismic districts of Etna (Baratta) 220 92 Map of the Ischian earthquakes of 1796, 1828, 1881 and 1883 (Mercalh) 223 93 Variation in earthquake-frequency before and after the erup- tion of the Usu-san (.Japan) in 1910 (Omori) . . . 224 94 Map of the Pendleton earth-shake of Nov. 25, 1905 . . 228 95 Portion of the meizoseismal area of the Californian earthquake of April 18, 1906 (Lawson) 231 96 Map of the principal after-shocks of the Inverness earthquake of 1901 236 97 Distribution of the centres of the principal after-shocks of the Inverness earthquake of 1901 ...... 237 98 Diagram illustrating the nature of the displacement that causes a simple earthquake ...... 238 99 Diagram illustrating the nature of the displacement that causes a twin earthquake . . . . . .241 100 Distribution of the audible after-shocks of the Mino-Owari earthquake of 1891 247 ABBREVIATIONS In addition to the usual abbreviations for well-known scientific journals, the following are used in the footnote references: Boll. Soc. Sis. Ital. Bollettino della Societa Sismologica Italiana. Bull. Eq. Inv. Com. Bulletin of the Imperial Earthquake Investigation Committee (Tokyo). Bull. Sets. Soc. Amer. Bulletin of the Seismological Society of America. DuUon. C. E. Button, The Charleston Earthquake of August 31st, 1886. Ninth Annual Report, U.S. Geological Survey, 1889, pp. 209-528. Lawson. The Californian Earthquake of April 18, 1906 (edited by A. C. Lawson), vol. 1 and atlas, 1908; vol. 2 (by H. F. Reid), 1910." Oldham. R. D. Oldham, Report on the Great Earthquake of 12th June, 1897. Mem. of the Geol. Surv. of India, vol. 29, 1899, pp. 1-379. Publ. Eq. Inv. Com. Publications of the Imperial Earthquake Investigation Conunittee in Foreign Languages (Tokyo). Seis. Journ . Seismological Journal of Japan . Trans. Seis. Soc. Japan. Transactions of the Seismological Society of Japan. CHAPTER I INTRODUCTION 1. All earthquake, in its widest sense, is a result of any sudden displacement of the earth's crust, either on or beneath its surface. The causes of the displacement may be natural or artificial, or partly natural and ijartly artificial, in their origin. The term is usually restricted, however, to movements of natural origin and to those which take place below the earth's surface; and it is in this sense that it will be used in this volume. Thus, the tremors caused by wind or sea-waves, by landslips or rockfalls, will not be re- garded as true earthquakes ; while those associated Avith volcanic operations or with the growth or shaping of the earth's crust will form the chief subjects of our inquiry. A disturbance of this nature is said to be seismic, and the science that deals with the phenomena and origin of such earth- quakes is called seismology. An earthquake felt at sea. and therefore consisting of mo\e- ments propagated through the sea, is called a seaquake. The initial displacement which results in an earthquake usually occurs at some depth below the surface, it may be of one or more miles. The displacement generates series of waves, which spread outwards with great velocit}' in all directions. As the waves pass any particle of rock, they cause that particle to move rapidly to and fro, and it is mainly this vibratory move- ment which produces the sensation of an earthquake. In some rare cases, however, the initial displacement, though in part deep-seated, is contimied up to the surface, and the vibratory motion is then complicated by the mass-displacement of the crust. 2. Each complete to-and-fro movement of an earth-particle is called a vibration. Let us suppose that a particle is executing vibrations along a straight line at right angles to Ali (Fig. 1). and that it can trace its course on a sheet of paper placed below it. If the paper D. M. 3. 1 INTRODUCTION [CH. at the same time be moved with uniform velocity in the direction BA, the path traced by the moving particle will take some form like that of the curve APCQB. In the simplest case, when the vibrations are not large, the motion is that known as simple harmonic motion, and the curve so drawn during one complete vibration is that corresponding to one complete period of the Fig. 1. Diagram of simple harmonic vibration. curve of sines. The portion APC or CQB corresponds to a semi- vibration. The time taken to execute a complete vibration (re- presented by the line AB) is called the period .of the vibration. The distance between the extreme positions, P and Q, of the particle measured along a line perpendicular to AB (that is, the sum of the distances PM and NQ) is called the range of the vibration. Half the range (that is, P3I or NQ) is called the amplitude of the vibration. The velocity of the particle is the rate at which it is changing its distance from some fixed point in its line of motion, say, from its position of rest in AB. The acceleration of the particle is the rate at which its velocity is changing*. * If tlie displacement at time t he a sin {M + e), where a, X and e are constants, the velocity and acceleration at the same moment are a\ sin ( X< + e + 7^ ) and a\^ sin {\t + e + tt). Thus, the amplitude or maximum displacement is a, the maximum velocity i] INTRODUCTION 3 The angle which the path of a vibrating particle at the surface of the earth makes with the horizontal plane through it is the angle of emergence. 3. While the term shock is often used as synonymous with earthquake, it should, strictly speaking, be confined to those vibrations which are sensible to the body of an observer. The vibrations which are so rapid that they affect the ear of an observer constitute the eaHhquake-sound. When the sound is heard without any attendant shock, it is called an earth-sound. The terms shock and sound are convenient, though there is no definite boundary between the two sensations, for the vibra- tions which produce the lowest sounds give rise to a quivering that is sensible to other parts of the body than the ear. Vibrations may be insensible to human beings either from the smallness of their amplitude or the lengths of their periods. Vibrations of the former class give rise to earth-iremnrs or micro- tremors, those of the latter class to earth-jJidsations. In French works, the following terms are sometimes used: seisms for earthquakes generally, microseisms for earth-tremors, macroseisms for sensible earthquakes, megaseisms for destructive earthquakes, and teleseisms for earthquakes of distant origin. In English, such terms are uncouth, but the corresponding- adjectives arc occasionally useful. 4. The place or region within which an earthquake originates is usually called (after Mallet) the seismic focus ov focus, occa- sionally the centre, centrum or hijpocentre. The chief objection to all such terms is the implication that the region in question is a point. Less objectionable, perhaps, is the word origin, meaning place of origin. No term, however, has obtained such cvu-rency as seismic focus or focus, and one or other will be used for the future, it being vmderstood that the focus is a region often of great extent. is «\, and the maximum acceleration rt.\-. If T be tlie period of a complete vibration, we have XT = 2ir or X= 2ir/T. Thus, the maximum velocity is 2irrt T, and the maximum acceleration ^ir-ajT-. It is shown in text- l)ooks of dynamics, or it may be deduced from the above expressions, that the maximum velocity is attained when the particle is passing tbrouph its position of rest (as at A, C or B), and the maximum acceleration when the particle is farthest from its position of rest (as at P or Q). 1—2 4 INTRODUCTION [ch. The area on the earth's surface vertically above the seismic focus — that is, the projection of the focus on the surface — is called the epicentre. The same objection applies to this term as to the words seismic focus and hypocentre. The epicentre is not a point, but an area of some, often of considerable, extent in at least one direction. 5. The intensity of an earthquake is proportional to the maximum acceleration of its vibrations, which is usually mea- sured in millimetres per second per second. An isoseismal line, or simply an isoseismal, is a line drawn through all places at which the intensity of a shock is the same. The meizoseismal area of an earthquake is the area within which the intensity is greatest. The term is somewhat indefinite and should perhaps be confined to the area included within the innermost isoseismal line*. The disturbed area of an earthquake is the district within which the shock is perceptible to the imaided senses. An isacoustic line is a line drawn through all places at which the same percentage of the total number of observers in it arc capable of hearing the earthquake-sound. The sound-area of an earthquake is the district within which the earthquake-sound is audible to some observers without in- strumental aid. 6. AVhen the epicentre of a great earthquake is submarine, the earthquake may be followed by a series of sea-waves. These are known as the seismic sea-waves, in Jaj^an as tsunamis. In popular writings, they are frequently called "tidal waves," an obvious misnomer, for such waves, though simulating tides of short period, are of entirely different origin. 7. A strong earthquake is sometimes, though not always, pre- ceded by a small number of slight shocks, and is invariably followed by a large number of a similar character. The great earthquake, with reference to the others, is called the principal shock or ^^n'/ici^pa? earthquake. The minor shocks are known as accessory shocks, those which occur before the * Mallet, to whom we are indebted for several of our terms, also defines a coseismal line as a line drawn through all points at which the same phase of the movement is felt at the same instant. The term is now seldom used. I] INTRODUCTION • 5 principal earthquake a.s fore-shocks, and those which occur after it as after-shocks. In districts near that in which an earthquake originates, shght shocks may be precipitated by the changes of stress intro- duced by the occurrence of the earthquake. These are known as sympathetic earthquakes. 8. Various classifications of earthquakes have been proposed and suitable names devised for them (see sect. 34). For the ])resent, it will be sufficient to refer to the two classes of volcanic and tectonic earthquakes. Volcanic earthquakes are those which precede, accompany, or follow, the operations of a volcanic eruption or are due to displacements within the mass of a ^■olcano. Tectonic earthquakes are the results of the growth of the earth's crust, that is, of the deformations to which the form of its surface-features is ultimately due *. * The present volume is concerned with the phenomena of earthquakes in peneral, and the description of in(iividual earthquakes hes beyond its range. It is important, liowcver, that such descriptions sliould be studied. Tiie most complete reports of recent earthquakes are those by C. E. Dutton on the Charleston earthquake of 1886 (Ann. Re]). U.S. Geol. Surv., vol. 9, l.S8!», |)p. 209-.)28), H. I). Oldham on the Assam earthquake of 1897 (Mem. (ii'ol. Surv. India, vol. 29, 1899. pp. 1-379), and A. C. Lawson (editor) on the C'alifornian eartliquake of 190(5 (Report of the State Earthquake Investi- oatiun Co)nrnission, vol. 1 and atlas, 1908; vol. 2, 1910). Frecpicnt references to tliese valuable rejjorts will be found in the following pages, the names of the authors and editor being given without the full titles of the books. Brief descriptions of various earthquakes, from the Neapolitan earthquake r>f 18.j7 to the .\ssam earthquake of 1897, arc given in C. Davison's Study of Recent Earthfiuakes (Contemporary Science Series), 190.j. CHAPTER II SEISMOGRAPHS 9. Of the instruments which have been designed for recording earthquakes, seismoscopes are intended merely to register the occurrence or the time of occurrence of an earthquake and they will not be considered in this chapter. The object oi seismometers, seismographs or tromometers is to record in detail at every moment of an earthquake the position of the ground relatively to its position of rest, so that, from the record or seismogram, we may determine the amplitude, direction of motion, maximum velocity and maximum acceleration of every single vibration*. A small movement of the ground is usvially composed of a displacement in some definite direction and a rotation about some definite line. Taking any three axes at right angles to one another — say, one vertical and the others horizontal — the dis- placement may be resolved into three comjoonent displacements * The literature dealing with the theory and construction of seismo- graphs is very extensive. The following books and memoirs may be men- tioned as among the more important : 1. Ewing, J. A. Earthquake measurement, yiem. Sci. Dtp., Tokyo Univ., No. 9, 1883, pp. 1-92. 2. Galitzin, Prince B. Vorlesungen iiber Seismometrie, 1914, pp. 1-538 (German translation, edited by O. Hecker). Galitzin's original memoirs on seismometry are published in the Comptes Rendus de la Commission Sismiqne Permanente (Petrograd), vols. 1-5, 1902-1912. 3. Knott, C. G. The Physics of Earthquake Phenomena (Oxford Univ. Press), 1908, pp. 48-89. 4. Marvin, C. F. A universal seismograph for horizontal motion and notes on the requirements that must be satisfied. Monthly Weather Rev. (U.S.A.), Nov. 1907, pp. 1-29. 5. Reid, H. F. Theory of the seismograph. Calif or nian Earthquake of April 18, 1906, vol. 2, 1910, pp. 143-190. 6. Walker, G. W. Modern Seismology (Longmans), 1913, pp. 1-36. 7. Wiechert, E. Theorie der automatischen Seismographen. Abhand. der kon. Gesell. Wissen. zu Gottingen, Math. Phys. Kl., vol. 2, 1903, pp. 1-128. Descriptions of numerous seismographs are given by R. Ehlert in Beitr. zur Geoph., vol. 3, 1896-1898, pp. 350-474; H. F. Reid in Bull. Seis. Soc. Amer., vol. 2, 1912, pp. 8-30; and G. W. Walker in Modern Seismology, pp. 16-20. CH. ii] SEISMOGRAPHS 7 along these lines, and the rotation into three component rota- tions about the same lines. A complete seismograph shonld therefore be capable of recording all six components, and no such instrument has yet been devised. It is probable, however, that, except near the epicentre, any movement of rotation is of little consequence, and thus the efforts of seismologists have been concentrated on the construction of instruments that will give a complete account of the displacements in three perpen- dicular directions, in other words, of instruments that will record the horizontal and vertical movements of the ground during an earthquake. It must be remembered, however, that an instru- ment designed for registering the horizontal motion in anj' direction will also be affected by a tilt or rotation of the ground, and that thus there may be some luiccrtainty whether the re- corded movement indicates a displacement only or a displace- ment complicated by a tilting of the ground. To describe the numerous seismographs which have been in- vented would require a volume as large as the present. All that can be attempted in this chapter is to give: (i) an outline of the principles on which the construction of seismographs is based, (ii) an account of a few instruments which are widely used in this country and Japan, and (iii) brief references to other in- struments, especially those designed by foreign workers. The movements of the ground during an earthquake may be of two kinds: (i) in great earthquakes, there may be a large displacement or lurch, either horizontal or vertical, or horizontal and vertical, the range of which may amount to 20 feet and more; and (ii) in all earthquakes, there is a vibratory motion of the ground, the period of the \ibrations \arying from a fraction of a second in near earthquakes to 20 or 30 seconds in distant earthquakes. The measurement of great lurches is beyond the scoj)e of seismographs*. For the record of the \ibratory move- * It may in some cases he made by a re-trian<;ulatioii of the district (sect. 84). To measure future movements alonji tlie San Ainheas fault in Cali- fornia, two series of concrete piers are sunk in the ground along lines at right angles to the fault, eacli series consisting of four piers, two on each side of the fault-line (Lawson, vol. 1, pp. 152-1.59). Vertical movements along the Hidgeway fault in Dorsetshire, if any should occur, will he determined by measuring the relative displacement of four brass castings lixed to the rock, two on each side of the fault (II. Darwin lieii. lirit. Ass., 19CM), pp. 119-120). 8 SEISMOGRAPHS [ch. ments, no single seismograph is sufficient. Different instruments are needed for the registration of near and of distant earth- quakes. The minute tremors which precede and accompany a volcanic eruption require a sensitive tremor-recorder or tromo- meter. In great earthquakes, the ordinary seismographs are usually damaged or thrown out of action, and a fourth form of seismograph, specially adapted for strong and vigorous motion, is necessary. The Essential Parts of a Seismograph 10. A seismograph consists of at least three parts: (i) the so-called steady mass, a certain point or line of which remains, or should remain, steady during the complicated movements of an earthquake; (ii) a frame or support, from which the steady mass is suspended, and which partakes in the movement of the ground; and (iii) the recorder, consisting of a lever or beam of light which magnifies the displacement, and a drum or plate (usually in continuous motion) on which the record is inscribed. In modern seismographs, there is also (iv) a damping device, the object of which is to check and control the oscillations which the steady mass may acquire during an earthquake. 11. The Steady Mass. Let AB (Fig. 2) represent the vertical axis of a cylinder pivoted about the horizontal line SS, and G the centre of gravity. If the point A be displaced along the line GA, the movement of the cylinder will be one of simple translation in the same direction. If the point A receive a small but sudden displacement in the direction (represented by the arrow) at right angles to the plane SAB, the movement of the cylinder will be one of rotation about a line //, parallel to SS, which meets AB in the point C. The distance AC is such that AC = k^JAG, where k is the radius of gyration of the cylinder about the line SS. The line // is called the instantaneous axis, and the point C the centre of jyercussion, with resjject to the line SS. Again, if the cylinder be pivoted about the line S'S' passing through B (Fig. 3), and if the point B receive a small displace- ment at right angles to the plane S'BA, the movement of the cylinder will be one of rotation about the parallel line /'/', which meets BA in the point C such that BC = k^jBG. II SEISMOGRAPHS Lastly, the cylinder may be pivoted about a vertical axis .S'.9 (Fig. 4), so that its axis AB is horizontal. If, then, the point A receive a small displacement in the direction at right angles S-' Fig. '2. Diagram illustratinward pull of the springs. ^Vhcn the bob is slightly displaced, the moment of the weight is nearly constant. To secure neutral equilibrium, the moment of the pull of the spring must also be made constant, and this is attained by Vifl. 11. Kwiny tliree-coniponent seismograph. attaching the lower ends E of the springs to a point below the rod AB, so that the horizontal distance of the springs from the fulcrum dimiiiislics or increases as the springs are lengthened or shortened. When the axis of supj^ort C receives a small vertical disjilacc- meiit, the instantaneous axis lies on the side of the centre of the bob M away from the axis of support. A cranked nuiltiplying lever L is attached to the frame AB at a point in the instan- 2—2 20 SEISMOGRAPHS [ch. taneoiis axis. The longer pai't of the lever is made of straw, and, at its lower end, carries a hinged marking pointer P, which rests lightly on the smoked glass plate. The diagram inscribed by this pointer is on a scale twice that of the vertical motion of the grovuid. 21. The positions of the two horizontal motion seismographs and the vertical motion seismograph are shown in Fig. 11, the former in front, the latter on the left, of the figure. They are arranged so that the three recording pointers are at different distances from the centre of the smoked glass plate. On the right is a clock, Avhich drives the glass plate at a uniform speed by a projecting friction-roller. The clock is started by an electro-magnetic detent, which acts as soon as an electric circuit is closed by a small seismoscope (shown on the right of Fig. 11 near the horizontal motion seismographs). This occiu's during the first preliminary tremors of an earthquake so that the plate begins to move before the principal vibrations arrive, and con- tinues in motion for two or three complete revolutions. Another clock (behind the revolving plate) is started by the same current as the driving-clock, the elapsed time at some subsequent moment giving the epoch at which the earthquake began. A small pointer projecting from this clock makes a small mark once a second on the glass plate*. 22. Gray-Milne Seismograph. The instrument described above is open to the objection that, once the plate has come to rest after an earthquake, no further shock can be registered until the starting apparatus has been re-set and a fresh smoked glass plate inserted. In the Gray-Milne seismograph, which resembles the Ewing seismograjoh in its principal mechanical details, this objection is overcome. The record is made in ink, by means of fine glass siphons, on a continuous strip of paper. Between earthquakes, this paper is driven at a uniform speed of from one-quarter of an inch to an inch per minute — a rate rapid enough to allow the time of occurrence to be determined with accuracy. With the first tremors of an earthquake, the speed is automatically increased to about 25, or even 50, inches per * T. Gray, Trans. Sets. Soc. Japan, vol. 3, 1881, pp. 137-139; Ewing, pp. 20-24, 49-51; also Proc. Roy. Soc, vol. 31, 1881, pp. 440-446; Trans. Sets. Soc. Japan, vol. 2, 1880, pp. 45-49; vol. 3, 1881, pp. 140-142. n] SEISMOGRAPHS 21 minute, the minutest details of the motion being thus distinctly rendered*. 23. Ewing Duplex-Pendulum Seismograph. The duplex-pen- dulum seismograph is a combination of a common pendulum, "which is stable, with an in^■crted pendulum, which is unstable, the masses and dimensions of the two pendulums being so arranged that the joint-system may be as neutral, or, rather, as feebly stable as may be desired. The mass M (Fig. 12) of the common pendulum is suspended by three wires, W, from the top of the case which contains the instrument. The mass .1/' of the inverted pendulum is sup- ported by a stout rod, which ends below in a conical point resting in a steel cup fixed to the bottom of ^^^ the case. A short vertical rod, ending in a spherical ball, projects upwards from the mass 3/' of the inverted pendulum and fits exactly into a cylindrical hole bored through the mass M. For neutral equili- brium, the lengths of the supporting wires of the mass M are arranged so that the circle of contact of the spherical ball and the cylindrical hole passes through the centre of percussion of both pendulums, namely, a little below the centre of gravity of the mass M and a little above that of the mass M'. By slightly lengthening the wires supporting the mass M. the necessary margin of stability is given to the joint -system. The multiplying lever consists of a light wooden rod L, the lower end of which terminates in a spherical ball fitting into the cylindrical hole through the mass M. \\ the upper end. a light arm of straw projects horizontally, with freedom of motion about a horizontal axis. The steel point at the end of the straw rests on a fixed smoked glass plate. The record thus shows the Fig. 12. Ewing duplex- pendulum seismograph. * T. Gray, Phil. Mas., vol. 2:?, 1KH7, pp. .•{.•):{-:5ration was 73 nun. In great eartlupuikes, even * S. Sekiya, Trans. Sets. Soc. Japan, vol. 12, 1888, pp. 83-100; F. Omori, Jimrn. Coll. Sci., Imp. Univ. Tokyo, vol. 11, 1899, p. 147, also Boll. Soc. Sis. Hal., vol. 2, 1890, p. 181, and Publ. Ei/. Inv. Com., No. 11, 1902, pp. 51-55, 0.3-04; V. Oniori and K. Ilirata. .lonrn. Coll. Sri.. Inij). Univ. Tokyo, vol. 11, 1899, pp. 191-193. 38 NATURE AND INTENSITY [ch. this amount is exceeded; but, in such cases, the overthrow of the seismographs prevents the registration of the exact amount. At Nagoya, the maximum range of the Mino-Owari earthquake of 1891 must have been about 223 mm. or 9 inches*. In the earthquakes recorded at the Hitotsubashi observatory (Tokyo) in 1887-1889, the average range of the largest vibration was -70 mm. in the horizontal, and -22 mm. in the vertical, component. In those recorded at the Hongo observatory (Tokyo) during the same years, the corresponding figures were -79 mm. and -22 mm. Of the 433 earthquakes recorded during the years 1885-1897 at the Central Meteorological observatory of Tokyo, the range was measured in 366 cases. In all but seven of these, the range of the largest vibration was less than 6 mm., and in about 60 per cent, of the total number, it did not exceed half a millimetre. In four earthquakes, the maximum range Avas considerable, namely, 22-8, 28-4, 41-0 and 76-0 mm.f 41. Maximum Acceleration of the Vibrations. In slight earth- quakes, the maximum acceleration is usually not more than 5 or 10 mm. per sec. per sec, the average for 64 earthquakes recorded at Hitotsubashi (Tokyo) being 20 mm. per sec. per sec. A value as high as 50 mm. per sec. per sec. is somewhat rare; and it is only in strong earthquakes that it exceeds 200 or 300 mm. per sec. per sec. In the semi-destructive Tokyo earth- quake of June 20, 1894, the maximum acceleration was 444 mm. per sec. per sec. at the Hongo observatory, and 900 mm. per sec. per sec. at that of Hitotsubashi {. The corresponding figures for certain destructive earthquakes are given in sect. 53. Intensity of Earthquake-Motion 42. Lower Limit of Sensible Motion. The least value of the maximum acceleration that is sensible to the unaided senses may be determined in two ways: (i) by the examination of the smallest values of the maximum acceleration of sensible earth- * The period of the largest vibrations at Nagoya seems to have been about 1-3 seconds, while the maximum acceleration, as determined by the overthrow of columns (sect. 43) was about 2600 mm. per sec. per sec. The formula / = -iw-a/T^ (sect. 2, footnote) gives the range of 2a equal to 223 mm.; see F. Omori, Publ. Eq. Inv. Com., No. 4, 1900, p. 17. t F. Omori, Publ. Eq. Inv. Com., No. 11, 1902, pp. 51-55, 94-95. X F. Omori, Boll. Soc. Sis. Ital., vol. 2, 1896, pp. 189-191. Ill] OF EARTHQUAKE-MOTION 39 quakes, deduced from seismographic records, and (ii) from the values determined at places which are known to be on or near the boundary of the disturbed area of an earthquake. (i) In his examination of the records of many sensible local earthquakes recorded at Moimt Tsukuba, Omori foimd that the range (or double amplitude) in the case of earthqiiakes unaccompanied by sound was seldom less than -013 mm., though it was equal to -01 mm. in 14 earthquakes accom- panied by sound. Taking the latter value as the lower limit of the range, and about one-tenth of a second as the period, the correspond- ing value of the maximum accelera- tion would be 17 mm. per sec. per sec. (ii) Omori has also examined the records of earthquakes registered in Tokyo in which that city lay on or close to the boundary of the disturbed area. In 23 earthquakes recorded at the obser- vatory of Hitotsubashi, the mean values of the range, period and maximimi acceleration were respcc- ti\cly -47 mm., -74 second and 17-0 mm. per sec. per sec. For 22 earth- quakes recorded at the observatory of Hongo, the corresponding values were ••35 mm., -64 second and 160 nmi. per sec. per sec. Thus Omori concludes that a maximum acceleration of 17 mm. per sec. per sec. is the lower limit of motion that is sensible without instrumental aid*. 43. Determination of the Maximum Acceleration by means of the Overthrow of Columns. In great earth(piakcs. such as that of Miiio-Owari (.Japan) in 1891, the range and maximum acceleration in the meizoseismal area are sufficient to put any ordinary seismograph out of action. In these cases, the over- throw of columns of known dimensions provides imder certain * Publ. Eq. Iiiv. Com., No. 11. 1!«)-_'. p. (;(); No. 22 A, 19()S, [jp. 37-39. my Fig. 22. OvcrturniniT of columns. 40 NATURE AND INTENSITY [ch. conditions a satisfactory substitute. Thus, let ABCD (Fig. 22) represent a cylindrical column resting on the ground at AB, y the height of the centre of gravity G above the ground, and X its horizontal distance from the edge B. If the period of the earthquake-motion be not very small compared Avith the period of rocking of the column, the column will move with the ground until the acceleration applied to it is sufficient to overthrow it. If, then, / be the maximum acceleration in the direction of the arrow, the least value of / that will overturn the column is giA^en by the equation mfij = 'Ifigx, the column being overthrown in the same direction as that in wdiich the movement of the ground at the time is taking place. The last equation is known as West's formula. Experiments to test the accuracy of West's formula have been made by Milne and Omori. Columns were placed on a shaking table, the movements of which Avere caused to simulate those of the ground during an earthquake by conforming to the law of simple harmonic motion. The exjoeriments show that the formula gives very nearly accurate results if (i) the period of the earthquake-motion be comparable with that of the columns when rocking, and if (ii) the amplitude of the earth's motion be not small. Thus, West's formula may be trusted to give accurate lower limits to the maximum acceleration experienced during a great earthquake*. An interesting confirmation of this result is furnished b)^ the use of columns of different materials, for West's formula depends on the dimensions only of the overturned body and not on its mass. Columns of brick and iron and Avooden boxes, Avith the same external dimensions, furnished nearly the same Aalues of the maximum acceleration f. * If / be the value of the maximum acceleration required to overthrow a column as given by West's formula and F the value determined by ex- periment, Omori finds that the mean value of the ratio f:F is 107: 1 {Publ. Eq. Inv. Com., No. 4, 1900, p. 136). t J. Milne, Trans. Seis. Soc. Japan, vol. 8, 1885, pp. 1-82; J. Milne and F. Omori, Seis. Journ., vol. 1, 1893, pp. 59-86; F. Omori, Seis. Journ., vol. 2, 1893, pp. 119-122; Boll. Soc. Sis. Ital., vol. 2, 1896, pp. 189-200; and Publ. Eq. Inv. Com., No. 4, 1900, pp. 69-141 , and No. 12, 1902, pp. 8-27 ; Bull. Eq. Inv. Com., vol. 4, no. 1, 1910, pp. 1-31. hi] of EARTHQUAKE-MOTIOX 41 44. Maximum Acceleration in Destructive Earthquakes. Omori has estimated the maxiniuni horizontal acceleration in several destructive earthquakes by observations on overturned bodies. In the Mino-Owari earthquake of 1891, he found the maximum acceleration to be 2500 mm. per sec. per sec. at Fukui, 2600 at Xagoya, 3000 at Gifu and Ogaki, 4000 at Kasamatsu, and more than 4300 at Iwakura and Komaki. In the Californian earthquake of 1906. the range of motion was about 4 inches and the period of vibration about 1 second, the corresponding- acceleration being aboiit 2000 mm. per sec. per sec. Omori estimated the maximum acceleration at Messina during the earthquake of 1908 at approximately the same figure. With these figures may be compared those which Oldham has gi\en for the maximimi horizontal acceleration during the Assam earthquake of 1897. He estimates it as 3000 mm. per sec. per sec. at Cherrapunji, 3600 at Gauhati. Shillong and Sylhet. and 4200 at Goalpara. At Shillong, Gauhati and other places in the epicentral tract, the actual value must have been far higher. At these places, stones Avere projected upwards, showing that the A'ertical component of the maxiniTmi accelera- tion must have been greater than that of gravity, which is 9000 mm. per sec. ])er see. * 45. Relations between the Nature of the Ground and the In- tensity of the Shock. In all earthquakes, the shock is felt more seserely on soft groinid than on hard compact rock. Milne, in his seismic survey of Tokyo made in the years 1884-1885, slK)wcd that the j)eriod of the more prominent earthquake- Aibrations was greater on soft than on comparati\ely hard ground, that the range of motion was greater in moderately strong earthquakes though not always in slight earthquakes, and that the maximum acceleration was also greater. Omori has also compared the records of two observatories in Tokyo. The observatory of Hongo is situated in the higher part of the city where the ground is of hard clay, that of Hitotsubashi * F. Omori, Boll. Sor. Sis. Itnl., vol. 2. 1896. pp. 192-197; Publ. Eq. Jiiv. Com., No. 4, 1900, pp. la-Hi: liull. Kq. Iiiv. Cow., vol. 1, 1907, p. 19, :m(l vol. :i, 1909, p. 40; Oldham. |)p. 78-79, 129-i:U. The nmxiinum accelera- tion (luring lateral vibrations of railway-carriajjes was foiiiul by Omori to be s(Jmetimes as much as 2000 niiu. per see. |)er see. (Hull. Ki/. Inv. Com., vol. 4., no. a, 1912, 1). 97). 42 NATURE AND INTENSITY [ch. stands on low ground consisting of very soft soil. Taking, first, the non-destructive earthquakes from Sep. 1887 to June 1889, the average period of the principal vibrations is -63 second at Hongo and -87 second at Hitotsubashi, the average maximum amplitude -56 and 1-07 mm., and the average maximum accelera- tion 20*4 and 24-9 mm. per sec. per sec. In the semi-destructive Tokyo earthquake of June 20, 1894, the period of the principal vibration was 1-3 seconds at Hongo and 1-7 seconds at Hitot- subashi, the maximum horizontal range 73-0 and 130-0 mm,, and the maximum acceleration 444 and 900 mm. per sec. per sec. 46. The most detailed study of the relation between the nature of the ground and the destructive power of the shock is that made by H. O. Wood of the damage wrought at San Francisco by the Californian earthquake of 1906. The city of San Francisco lies between about 1 and 9| miles east of the great fault, the movement along which gave rise to the earthquake (sect. 84). On the whole, the intensity of the shock decreased with in- creasing distance from the fault, but it was subject to many variations evidently connected with the nature of the ground. The shock was slightest, resulting in the occasional fall of chimneys, in a few small areas, which are invariably those occupied by hard rock (sandstone, chert, etc.) with a level surface. A higher degree of intensity, corresponding to general, but not universal, fall of chimneys, with cracks in masonry and brickwork, marks ground consisting of hard rock with an inclined surface or hard rock with a thin coating of soil. On thick beds of naturally formed alluvium, old and well-com- pacted, brickwork and masonry were badly cracked, some gables were thrown down, and chimneys generally were destroyed. The worst damage occurred on newly made land, especially on that filling up a marsh or creek. Here, brick and frame buildings generally collapsed, the surface of the ground was thrown into broad undulations, sewers and water-mains were broken*. * J. Milne, Trans. Seis. Soc. Japan, vol. 10, 1887, pp. 1-36; vol. 13, 1890, pp. 41-89; C. Davison, The Hereford Earthquake of Dec. 17, 1896 (1899), pp. 276-278; F. Omori, Publ. Eq. Inv. Com., No. 11, 1902, pp. 57-58, and Boll. Soc. Sis. Ital., vol. 2, 1896, p. 191 ; H. O. Wood, The Californian Earth- quake of April 18, 1906 (Lawson), vol. 1, 1908, pp. 220-245. M. Baratta gives roughly the same sequence as H. O. Wood for the Messina earthquake of 1908 in Catastrophe Sismica Calabro-Messinese, 28 dicembre 1908, 1910, p. 230. in] OF EARTHQUAKE-MOTION 43 47. Comparison between the Motion on the Surface and in Pits. To the unaided senses, there is a perceptible difference between the intensities of an earthquake at the surface and in mines, a shock which is strong on the surface being hardly felt or not felt at all at some depth below. In the Hereford earthquake of 1896 and the Derby earthquake of 1903, the distance from the epicentre of the farthest mine in which the shock Avas felt was only one-third of the mean radius of the isoseismal 4 (sect. 51). In the meizoseismal area of the Riviera earthquake of 1887, the shock was very weak or not felt at all in the tunnels of the Nice to Genoa railway, and none of the tunnels was damaged in the slightest degree. There is also a marked difference in the strength of a shock in slight hollows or excavations at the surface and on the adjoining ground. For instance, in the central tract of the Mino-Owari earthquake of 1891. the railway-lines were every- where more or less disturbed except in small cuttings. Even if the cuttings were not more than 20 or 50 feet in depth, the rails and sleepers were unmoved. At Dharmsala, during the Kangra earthquake of 1905, one house surrounded on several sides by higher spurs and ridges, was spared while the destruction of Ijuildings on elevated ground was either total or nearly so*. 48. The reason for this comparati\e immunity from the shock in hollows and mines is furnished by some interesting experiments of Sekiya and Omori made at Tokyo in the years 1887 to 1889. The records of two similar seismographs, one at the bottom of a pit 18 feet deep, the other on the surface within a few yards from the pit, were compared for thirty earthquakes, of which three were severe and the rest slight. For the latter, it was' found that the average amplitude, maximum velocity and maximimi acceleration differed but little on the surface and in the pit, each being slightly greater on the surface. In the large undulations of the severe earthquakes, the three elements again differed but slightly, the greater magnitude being in each case on the surface. It was in the rijijiles of these earthquakes that * C. Davison, Ilerefonl Kaii/K/ioihc, pp. 27H-28(), and Quart. Jaurii. (icol. Soc, vol. (JO, 1904., i)p. '227-22H; A. IsscI, lioll. del li. Com. Geol. d Jlalia. anno 1887, pp. 117-1*20; J. Miliic, Sei.s. Journ., vol. 1, 1893, p. 133; C. S. Middlemiss, Meni. Geol. Surv. India, vol. 38, 1910, p. 21. 44 NATURE AND INTENSITY [ch. the greatest difference was manifested, the amphtude being about twice, the maximum velocity three times, and the maximum acceleration five times, as great at the surface as in the pit. Owing to their much shorter period, the ripples at the surface have a maximum acceleration from five to ten times as great as that of the large undulations. Thus it would seem from these observations that the ripples are in great part smoothed away in the pit and that there should be much less destructive action in houses with foundations rising from deep pits than in those built on the free surface*. Scales of Seismic Intensity 49. Uses of Scales of Intensity. The determination of the maximum acceleration by means of overthrown columns and fractured walls is possible only in strong earthquakes and in their central areas. Nor, on account of their cost, can seismo- graphs be widely used. Various arbitrary scales of seismic in- tensity have therefore been suggested, their chief objects being (i) to compare the intensities of different earthquakes, and (ii) to trace, by means of isoseismal lines, the variation of intensity in a shock throughout its disturbed area. A good scale may, indeed, be more useful for the latter purpose than accurately constructed seismographs, for it enables us to obtain a large number of observations of the intensity from within a limited area. 50. Conditions to be fulfilled by Scales of Intensity. The following conditions should be fulfilled by any satisfactory scale of intensity : (i) Each degree shovdd represent a constant intensity. It should depend on the mechanical effects of the shock, and not on personal impressions which may vary in different coimtries and with different observers in the same country. (ii) Each degree should consist of one test only, imless the exact equivalence of two or more tests has been determined. * Trans. Scis. Soc. Japan, vol. 16, 1892, pp. 19745. Observations of a similar, but less detailed, character were made by Milne two or three years earlier (Trans. Seis. Soc. Japan, vol. 10, 1887, pp. 2.5-26, 36). The observa- tions described above show that the deep cuttings through which the Panama Canal passes may to some extent be immime from the effects of local earthquakes. Ill] OF EARTHQUAKE-MOTION 45 (iii) The degrees of a scale should be so far apart that an intelligent observer should have no difficulty in distinguishing between the tests of successive degrees ; and yet they should be close enough to be applicable to the earthquakes of any country and to shocks of all degrees of strength. The scales most widely used are the Rossi-Forel scale and the Mercalli scale (sects. 51 and 52). The latter is suitable for strong earthquakes and is adopted in Italy as the standard scale; the former is slightly better adapted to earthquakes of moderate strength and is widely used in other seismic countries. It will be noticed that neither scale satisfies the first and second of the above conditions. Both depend to some extent on the effect of the shock on the observer, and in each the average number of tests to a degree is three. 51. Rossi-Forel Scale (1883). (1) Recorded by a single seis- mograph, or by some seismogra[)hs of the same pattern, but not by several seismographs of different kinds; the shock felt by an experienced observer. (2) Recorded by seismographs of different kinds; felt by a small number of persons at rest. (3) Felt by several persons at rest; strong enough for the duration or direction to be appreciable. (i) Felt by several persons in motion; disturbance of move- able objects, doors, windows, creaking of fioors. (5) Felt generally by everyone; disturbance of furniture and beds ; ringing of some bells. (6) General awakening of those asleep; general ringing of bells; oscillation of chandeliers, stopping of clocks; visible dis- turbance of trees and shrubs ; some startled persons leave their dwellings. (7) Overthrow of moveable objects, fall of plaster, ringing of church-bells, general panic, without damage to buildings. (8) Fall of chimneys, cracks in the walls of buildings. (9) Partial or total destruction of some buildings. (10) Great disasters, ruins, disturbance of strata, fissures in the earth's crust, rock-falls from mountains*. * Arch, dcs Sci. pfii/s. ct itut., vol. 11, 1881, pp. 148-149. A siinplilicd form of this scale (witli only one tost for each def^ree), used in the investi- ^'ation of British earthquakes, is pivcn in Phil. Mag., vol. 50, 1000, p. 51; Geogr. Jotirn., vol. 46, 1915, pp. ;160-361. 46 NATURE AND INTENSITY [ch. 52. Mercalli Scale, (l) Instrumental shock, that is, noted by seismic instruments only. (2) Very slight, felt only by a few persons in conditions of perfect quiet, especially on the upper floors of houses, or by many sensitive and nervous persons. (3) Slight, felt by several persons, but by few relatively to the number of inhabitants in a given place; said by them to have been hardly felt, without causing any alarm, and in general without their recognising it was an earthquake until it was known that others bad felt it. (4) Sensible or moderate, not felt generally, but felt by many persons indoors, though by few on the ground-floor, without causing any alarm, but with shaking of fastenings, crystals, creaking of floors, and slight oscillation of suspended objects. (5) Rather strong, felt generally indoors, bvit by few outside, with waking of those asleep, with alarm of some persons, rattling of doors, ringing of bells, rather large oscillation of suspended objects, stopping of clocks. (6) Strong, felt by everyone indoors, and by many with alarm and flight into the open air; fall of objects in houses, fall of plaster, with some cracks in badly-built houses. (7) Very strong, felt with general alarm and flight from houses, sensible also out-of-doors ; ringing of church-bells, fall of chimney- pots and tiles; cracks in numerous buildings, but generally slight. (8) Ruinous, felt with great alarm, partial ruin of some houses, and frequent and considerable cracks in others; without loss of life, or only with a few isolated cases of personal injury. (9) Disastrous, with complete or nearly complete ruin of some houses and serious cracks in many others, so as to render them uninhabitable ; a few lives lost in different parts of populous places. (10) Very disastrous, with ruin of many buildings and great loss of life, cracks in the grovmd, landslips from mountains, etc.* 53. Absolute Scales of Intensity. The defects of arbitrarv scales of intensity are so obvious that some seismologists have attempted to replace them b3^ absolute scales, the maximvmi intensity in each degree being expressed by the corresponding- acceleration in millimetres per second per second. * BoU. Soc. Sis. Ital, vol. 8, 1902, pp. 184-191. Ill] OF EARTHQUAKE-MOTIOX 47 The earliest scale of this kind is that proposed by Omori, which is applicable only to strong earthquakes, the maximum accelerations corresponding to successive degrees being 300, 900, 1200, 2000, 2500, 4000, and much more than 4000, mm. per sec. per sec. A few years later, Cancani suggested a scale for shocks of all intensities, consisting of twelve degrees, the maxi- mum intensities for each in mm. per sec. per sec, being 2-5, 5, Fig. 23. Construction of isoseismal lines. 10, 25, 50, 100, 250, 500, 1000, 2500, 5000, and 10,000. McAdic has recently brought forward a third absolute scale, which is the same as Cancani's except that the first three degrees of the latter are grouped as one. At the present time, however, we have no means for obtaining mmierous determinations of the maximum acceleration when it is less than 200 or 300 mm. per sec. per sec, and thus Omori's scale is the only one that is now of practical value*. * F. Oniori, Publ. Eq. Inv. Com., No. 4, 1!»00, pp. It, 137-Ul; A. Can- cani, Verb, der II. intern, seis. Konf., 1904; A. McAdie, Bull. Seis. Soc. Amer., vol. .5, 1915, p. I'i.*}. 48 NATURE AND INTENSITY IsosEisMAL Lines and Disturbed Area [CH. 54. Construction of Isoseismal Lines. The accompanying sketch-map (Fig. 23) ilhistrates the method of drawing iso- seismal lines. It represents the central area, bounded by the isoseismal line of intensity 8, of the Hereford earthquake of Fig. 24. Isoseismal lines of the Charleston earthquake of 1886. 1896. At all places marked by black dots, the intensity was not less than 8 (Rossi-Forel scale), that is, chimneys were thrown down or walls cracked. Places at which no damage was re- ported are indicated by small circles. The isoseismal line is then drawn so as to include all the former places and as far as possible to exclude the latter. For a lower degree of intensity, say 6, the observations may provide intensities which are certainly 6, Ill OF EARTHQUAKE-MOTION 49 probably 6, possiblj'^ 6, and certainly less than 6. The line would then be drawn so as to include places of the first class, many or most of those of the second, some of those of the third, and none of those of the fourth class*. 55. Forms of Isoseismal Lines. Several examples of the forms of isoseismal lines are giveii in the maps in this vohnne. As examples of their form in the comparatively slight earthquakes of Great Britain may be mentioned those of the Helston earth- quake of 1898 (Fig. 30) and the Derby earthquakes of 1903 and 1904 (Figs. 28 and 51). The isoseismal lines of the Charleston earthquake of 1886 are reproduced in Fig. 24 f. Fig. 26 shows the isoseismal lines of the Mino-Owari earthquake of 1891, in which Omori's absolute scale of intensity is used. The difference in the forms of the Charleston and British isoseismals — the undulations of the one and the regularity of the others — should be noticed. The difference is chiefly due to the use of several tests for each degree of the Rossi-Forel scale in the Charleston earthcpiake and of one test only for each degree in the British earthquakes; and partly to the very large munber of observations available in the British earthquakes. Probably, it is also due in some measure to the wide variations in the nature and form of the ground traversed by the Charleston earth-waves. 56. Magnitude of the Disturbed Area. The disturbed areas of earthquakes range between wide limits. That thearea depends to a great extent on the intensity of the shock is clear from the follow- ing table, in which is given the disturbed areain British earthquakes (1889-1909) for each degree of the Rossi-Forel scale from 8 to 3: Intensity Disturbed area in sq. miles Max. Min. Average 8 7 6 5 4 3 98000 63600 3100 3000 1130 219 33000 KMM) 74 90 28 81i 65900 24500 1200 850 260 126 * Kor further details regardiiiff tlie construction of isoseismal lines, see niilr. zur (ienph., vol. 9, 1908, i)p. 21+-218. t Dutton, plate 29. D. M.S. 4 50 NATURE AND INTENSITY [ch. From the wide difference between the maximum and minimum areas for each degree, it is evident that other factors, besides the intensit}^ govern the extent of the disturbed area. In volcanic earthquakes, strong enough to ruin the epicentral villages, the area ranges from 50 to about 1000 sq. miles (sect. 227). As a rule, a destructive tectonic earthquake is felt over an area of one-quarter to one-third of a million square miles, as, for example, over 219,000 sq. miles in the Riviera earthquake of 1887, 230,000 sq. miles in the Bengal earthquake of 1885, about 250,000 sq. miles in the Cachar earthquake of 1869, 330,000 sq. miles in the Mino-Owari earthquake of 1891, and about 373,000 sq. miles in the Californian earthquake of 1906. Occasionally, much larger areas are shaken. For instance, the Assam earthquake of 1897 disturbed about If million sq. miles, and the Kangra earthquake of 1905 nearly 2 million sq. miles. The largest known disturbed area is that of the Charleston earthquake of 1886, which covered about 2,800,000 sq. miles. In this case, however, the earthquake was not one of the first order of magnitude, and its extensive disturbed area (bounded by an isoseismal of intensity 2, Fig. 24) was mainly due to its occurrence within an area occupied by civilised races *. Direction of Earthquake-Motion 57. Direction as revealed by Seismographs. The direction of motion during an earthquake is rarely rectilinear. In successive vibrations, the movement may take place in all azimuths. An example, perhaps an extreme example, of this varied movement is given by Sekiya, who represented the motion of an earth- particle during the Japanese earthquake of Jan. 15, 1887, by a model reproduced from the three components of the motion (see also Fig. 21). Another example, perhaps also an extreme one, is that of the Tokyo earthquake of June 20, 1894 (Fig. 19). Though the direction of motion, as usual, changed during this earthquake, the maximum horizontal motion was directed towards S. 70° W., and the principal movements both before and after were also in the same or opposite direction. The epicentre was situated to the east of Tokyo. * Geo/. Mog., 1910, p. 412. Ill] OF EARTHQUAKE-MOTION 51 The direction of the maximum horizontal motion has been compared with that of the epicentre from the observing station in a number of Japanese earthquakes recorded at Miyako (1896- 1898) and Tokyo (1887-1889). When the earthquakes are strong and the epicentres at no great distance, there is in many cases a rough agreement between the two directions. On the other hand, when the earthquakes are weak, there seems to be no prevaiHng direction discernible in the vibrations*. 58. Mean Direction determined from the Overthrow of Columns. Some interesting observations have been made by Omori on the directions in which colimms of various forms were overthrown by the Tokyo earthquake of 1894'. In Tokyo, he W-«^=^^^'=^^ ^ — _> — E Fig. 25. Direction-rose for the Tokyo earthquake of .June 20, 1894. measured the directions in which 245 ishidoro (stone lanterns in gardens) and other bodies fell. Of these bodies, 144 were ishidoro with circular bases. The colimms fell in various directions, the numbers within successive angles of 15° being represented in the diagram in Fig. 25. The mean of the 245 directions was S. 71° W.-X. 71° E., which coincides almost exactly with the direction of the maximum movement recorded in the last section. Of a total number of columns, 159 (or 65 per cent.) fell within the two quadrants adjacent to the direction of maximum movement, and 86 (or 35 per cent.) in the other (piadrants. Similar observations were made by Omori at a number of * S. Sckiya, Trans. Seis. Sdc. Japan, vol. 11 , 1887, pp. 175-177 ; S. Sekiya and F. Omori, Jonrti. Cull. Sci., Imp. Univ. Tokyo, vol.57, 1894, |)p. 1-4; F. Omori, Pt4bl. Eq. Inv. Com., No. 11, 1902, p. Mi\ F. Omori and K. Hirata, ./ourii. Coll. Sci., Imp. Univ. Tokyo, vol. 11, 1899, pp. 193-194. Sekiyas model, referred to above, is reproduced in Nature, vol. 37, 1888, p. 297. 4—2 52 NATURE AND INTENSITY [CH. places within and near the meizoseismal area of the Mino-Owari earthquake of 1891. The area within the dotted hne in Fig. 26 represents the zone of extreme violence. The two curves, marked 800 and 2000, are the isoseismal lines referred to in sect. 55. The mean directions of motion in different places are indicated by the short arrows, the arrow-heads pointing in the direction •. -. -s V ^^.-'' \ •■. •. ^. ^ ■^ — " \ •• •. \ 800 \ ■• -. ^ \ ■. •. \ ^ ■• •. \ \ •. ■ \ '• ■-. \ •. ■■ \ ■. \ I ': 1 : / ; /■■■• \ \ — *—~\y^ Nagoy a / ^ 1 / \ j^f^^'"'^'^^ I 1 / r '\~'^ 1 I I , / / / \ \ 1 ■* '^-■;^> ^ /{ '^ -^Op ■I \ J* T '^yu - — Fig. 26. Map of the directions of the shock and of the absolute isoseismal lines of the Mino-Owari earthquake of 1891. in which the majority of bodies in each place Avere overthrown. It will be noticed that, in nearly every case, the mean direction of motion was at right angles to, and directed towards, the meizoseismal zone*. 59. Mean Direction determined by Personal Impressions. Ob- servations on the direction of a shock are greatly influenced by the direction of the principal walls of the house. In four strong * F. Omori, Publ. Eq. Inv. Com., No. 4, 1900, pp. 17-22, 25-33; Boll. Soc. Sis. Hal., vol. 2. 1896, pp. 184-187. Ill] OF EARTHQUAKE-MOTION 53 British earthquakes, the apparent direction of the shock agreed with the direction of the principal walls of the house in 70 per cent, of the observations, the mean deviation of the apparent direction from the direction of the walls being only 9^°. Now, the mean of a large mniiber of observations in any place on the apparent direction is found to coincide very nearly with the direction of the line joining the place to the epicentre. For example, in the Hereford earthquake of 1896, the deviation of the mean apparent direction from the direction of the epicentre was 2° in Birmingham and London; while, in the Derby earth- quake of 1904, the two directions were coincident at Derby. The explanation no doubt is that the sense of direction is most a])])arent in houses in which the principal walls are parallel or perpendicular to the true direction of the shock*. Duration of Earthquake-Motion 60. Duration of the Preliminary Tremor. With some rare exceptions, the preliminary tremor is present in every seismo- gram. Its duration, as Omori has shown, depends, not on the strength of the shock, but solely on the distance of the station from the origin. For instance, the preliminary tremor of the Mino-Owari earthquake of 1891 lasted 2 seconds at Gifu, while the average duration in five of the after-shocks was also 2 seconds. On the other hand, the duration for the principal earthquake was 14 seconds at Osaka (distant 87 miles) and 37 seconds at 'i'okyo (179 miles). In Fig. 27, the relation between the distance of the origin and the duration of the preliminary tremor is represented for various .Japanese earthquakes from 1891 to 1900. The distance (x) is measured in kilometres, the duration (y) in seconds. The cor- r(sj)onding points are grouped close to, and on either side of, I lie straight line in the figure, the equation of Avhich Omori linrls to be ^ _ 7.27^ + 38. In these earthquakes, the distance of the epicentre lies between 70 and 900 kms. For earthquakes in which the distance lies between 50 and 200 kms., Omori gives the formula X = 6-86t/ + 8-1, * C. Davison, licilr. zitr Geoph., vol. 8, 190(5. i)p. 7:}-74; vol.9, 1908, \>\>. 219-220. 54 NATURE AND INTENSITY [CH. and, for ordinary earthquakes with an origin less than 1000 knis. distant, ^ _. 7.491/* 61. Duration of the Earthquake. To the unaided senses, the total duration of an earthquake varies from one or a few seconds for a weak or moderately strong earthquake to 3 or even 4 minutes for one of destructive violence. In the stronger British earthquakes, the mean duration ranges from 4*0 seconds for the Stafford earthquake of 1916 to 10-5 seconds for the Hereford earthquake of 1896. In great earthquakes, single ecs 120 y^ y^ 100 y • 80 / 60 y 40 y^' > y. \ 20 n ^ ■. i 100 200 300 400 500 600 700 800 900 kms. Fig. 27. Relation between the duration of the preliminary tremor and the distance of the station from the origin. estimates of the duration amount to 70 seconds in the Charleston earthquake of 1886, 3| minutes in the Californian earthquake of 1906, and as much as 5 minutes in the Assam earthquake of 1897; but, in such cases, it is always possible that the estimates include the duration of closely succeeding after-shocks. When the movement is registered by seismographs, the dura- tion is much greater. For instance, the records of a Gray-Milne seismograph at Miyako (Japan) from 1896 to 1898, give total durations ranging from 8-5 to 200 seconds. The following table * F. Omori, Puhl. Eq. Inv. Com., No. 13, 1903, pp. 88-91; Bull. Eq. Inv. Com..., vol. 1, 1907, pp. 145-154; vol. 2, 1908, pp. 144-147; vol. 6, 1914, p. 238; and vol. 9, 1918, pp. 33-39. Ill] OF EARTHQUAKE-MOTION 55 gives the durations of the prehminary tremor, principal portion and end-portion in some of these earthquakes, the principal portion being clearly defined only when the epicentre is not more than 200 kms. distant. Duration in seconds Distance of epicentre in kms. Eartliqiuike Preliminary Principal End-por- tremor portion tion 1896. Aug. 31 (1) 100 10 4 59 „ (2) 100 52 26 69 1897. Mar. 27 70 4-2 24 52 Apr. 30 60 2-6 0-7 39 June 18 90 8-9 9 32 Aug. 23 110 11 9-3 80 Oct. 2 150 12 5 13G 84 Dec. 23 70 3-6 3-6 28 1898. Apr. 23 200 13 1-9 105 Average — 7-9 10-2 61 Of 487 earthquakes recorded in 1905 by an Omori horizontal tremor recorder at Mount Tsukuba (Japan), the total durations ^'aried from 5 to 275 minutes, 64 per cent, of the total number having diu'ations of less than 1 minute, 28 per cent, between 1 and 2 minutes, 4 per cent, between 2 and 3 minutes, 2 per cent, more than 3 minutes, the durations of the rest being inde- terminate. Again, the apparent duration of an earthquake depends to a great extent on the sensitiveness of the seismograph employed. P'or instance, the duration of one of the above earthquakes was 2 minutes according to the record of the Gray-Milne seismograph at Miyako, and 2 hours as registered by an Omori horizontal pendulum at Tokyo; the former instrument responding to the quicker vibrations only and not to the slow undulations forming the end-portion of the movement*. * F. Omori and K. Ilirata, Journ. Coll. .S'c/.. Imp. Univ. Tokyo, vol. 11, 1899, pp. 189-191 and table; F. Omori, Publ. Eq. Inv. Com., No. 22 .\, 1908. pp. 1-39. CHAPTER IV THE SOUND-PHENOMENA OF EARTHQUAKES 62. The vibrations which are perceptible as sound have a wide range of periods, the lowest audible note being produced by about 30 vibrations per second, and the highest by about 70,000. Both limits are, however, subject to variation, in different persons as well as in different races*. 63. General Character of the Sound. The sound Avhich accom- panies an earthquake is usually a low heavy rumbling noise, a deep booming or low moaning, a grating roaring or a crushing- grinding noise. Sometimes, very different sounds are heard like that of a roaring wind or of a chimney on fire. The chief characteristic of the sound is its extraordinary depth. It is almost too low to be heard. It is described as a rumble that can be felt. The impression of great depth is also conveyed by the frequent use of the word "heavy," in such comparisons as heavy peals of thunder, heavy gusts of wind, or the heavy rumbling of sea-waves in a cave. But no evidence of its extreme lowness is so decisive as the fact that the sound is heard by some observers and not by others. To one the sound * The following papers may be consulted, on the general phenomena of earthquake-sounds : 1. Davison, C. (1). On earthquake-sounds. Phil. Mag., vol. 49, 1900, pp. 31-70. 2. (2). The sound-phenomena of British earthquakes. Beitr. zur Geoph., vol. 12, 1913, pp. 485-527; and on brontides: 3. Alippi, T. (1). I mist-poeffeurs calabresi. Boll. Soc. Sis. Ital., vol. 7. 1901, pp. 9-22. 4. (2). Di un fenomeno acustico della terra o delF atmosfera. Ibid. vol. 12, 1907, pp. 1-42. 5. (3). Nuovo contributo all" inchiesta sui "Brontidi." Ibid. vol. 15, 1911, pp. 65-77. 6. Cancani, A. (1). Barisal-guns, mist-poeffeurs, marina. Ibid. vol. 3, 1897, pp. 222-234. 7. (2). Rombi sismici. 76iV/. vol. 7, 1901, pp. 23-47. 8. Van den Broeck, E. Un phenomene mysterieux de la physique du globe, del et Terre, vol. 16, 1895, and vol. 17, 1896. CH. IV] SOUND-PHENOMENA OF EARTHQUAKES 57 seems like that of a loud explosion, to another in the same place like distant thunder, to a third the shock appears un- accompanied by sound. 64. Types of Earthquake-Sound. Occasionally, the sound which accompanies an earthquake is supposed to be unlike any known soimd. As a rule, however, it is compared to one of the tyjjcs of the following scale, known as the Davison sound-scale: 1. Waggons, carriages, traction-engines or trains passing, generally very rapidly, on hard ground, over a bridge or through a tunnel; the dragging of heavy boxes or furniture over the floor. 2. Thunder, a loud clap or heavy peal, but most often distant thimder. 3. Wind, a moaning-, roaring- or rough strong wind : the rising of the wind, a heavy wind pressing against the house, the liowling of wind in a gap or a chimney, a chimney on fire, etc. 4. Loads of stones, etc., falhng, such as the tipping of a load of coals or bricks. 5. Fall of heavy bodies, the banging of a door, the blow of a wave on the seashore. 0. Explosions, distant blasting, the boom of a distant heavy gun. 7. Miscellaneous, such as the trampling of many animals, an inniiense covey of partridges on the wing, the roar of a waterfall, a low pedal note on the organ, and the rending or settling to- gether of huge masses of rock. In strong British earthquakes (those in which the isoseismal 4 includes an area of more than 5000 sq. miles), 46 per cent, of the observers compare the sound to passing waggons, etc., 24 per cent, to thunder. 11 per cent, to Avind. 5 per cent, to loads of stones falling, 3 per cent, to the fall of a heavy body, 7 per cent, to explosions, and 5 per cent, to miscellaneous sounds. In very slight earthquakes (disturbing areas of less than 60 sq. miles), tile figures are different: 9 per cent, compare the sound to ])assing waggons, etc., 11 per cent, to thunder, 2 per cent, to wind, 9 per cent, to loads of stones falling, 25 per cent, to the tall of a heavy body, 42 per cent, to explosions, and 2 jxr cent, to miscellaneous sounds. \s a rule, the sound adheres throiighout to one of the types mentioned abo\ c. and \ arics. if at all, only in intensity, be- 58 SOUND-PHENOMENA OF EARTHQUAKES [ch. coming gradually louder and then dying away. The places at which changes of type are observed are situated for the most part within a district closely surrounding the epicentre, and the changes take place at the moment when the sound is loudest and the strongest vibrations are felt. Occasionally, a loud ex- plosive crash, like that of heavy blasting, is heard at this moment, or a sound like that of rushing wind may merge into that of a heavily loaded traction-engine passing at a rapid rate *. 65. Inaudibility of the Sound. The inaudibility of the sound is either total or partial. Partial inaudibility may consist in the suspension of all sound during part only of the time while it is heard by others, or in the suppression of some vibrations only, so that observers in the same place may refer the sound to different types f or that some may hear the deep explosive crashes of which others are vmcon scions. In British earthquakes, the sound is heard almost invariably, the omission of reference to the sound in exceptional cases being probably accidental. In strong earthquakes, 83 per cent, of all the observers hear the sound. In slight earthquakes (those in which the isoseismal 4 includes areas of less than 1000 sq. miles), the percentage rises to 97. For this higher percentage, there are two reasons: (i) the sound in slight earthquakes is usually a more prominent feature than the shock, and (ii) in strong earth- quakes, the sound-area is less extensive than the disturbed area, M'hile in slight earthquakes the two areas are approximately coincident. The inaudibility of the sound to some observers is often attributed to inattention. When the sound appears to some as a deafening noise or as louder than the loudest thunder, such an explanation must obviously fail. As this is the case, how- ever, near the epicentre only, a more decisive test is furnished, in the case of earthquakes Avhich occur at night, by the audi- bility of that part of the sound before the vibrations begin to be felt. In the Hereford earthquake of 1896 (5.32 a.m.), the * Davison (2), pp. 488-497. f For instance, at Birmingham during the Hereford earthquake of 1896, 36 per cent, of the observers refer to passing waggons, 18 per cent, to thunder, 18 per cent, to wind, 4 per cent, to loads of stones falhng, 6 per cent, to the fall of a heavy body, 11 per cent, to explosions, and 7 per cent, to miscellaneous sounds. IV] SOUND-PHENOMENA OF EARTHQUAKES 59 fore-sound was heard by 72 per cent, of those who were awake and 74 per cent, of those who were asleep; in the Inverness earthquake of 1901 (1.24 a.m.), the corresponding figures are 72 and 72; and in the Doncaster earthquake of 1905 (1.37 a.m.) 78 and 67. Moreover, in the Hereford earthquake, the per- centages within the isoseismal 8 were 78 for those awake and 75 for those asleep; at a considerable distance, in the zone between the isoseismals 6 and 5, the corresponding figures were 61 and 60. Thus, roughly, the preliminary sound is equally audible to all observers, whether awake or asleep, and therefore the inaudibility of the sound to some cannot be due to in- attention. It can only be explained on the supposition that the sound-vibrations of an earthquake are in the immediate neighbourhood of the lower limit of audibility, and that this limit varies in different persons, so that some may be deaf to such low sounds though by no means deaf to ordinary noises. The partial inaudibility of the sound is no doubt due in part to the same cause, especially as regards the deep explosive crashes heard by some, and not by others, when the shock is strongest. The temporary cessation (to some) of all sound while the shock is felt may be due to this cause, partly also to fatigue*. 66. Variation in Audibility throughout the Sound-Area. In strong British earthquakes, it is usually possible to calculate the percentage of audibility in five zones bounded by successive pairs of isoseismals. The average percentages for such earth- quakes are 97 within the central isoseismal, and, in successive zones, 94, 88, 69 and 60. There is thus, as we should expect, a decline in audibility as the distance from the origin increases, but the rate of decline is at first slow, and afterwards more rapid, especially near the boundary of the sound-area. From this rapid decline, we may infer that the lower limit of audibility does not vary much in different observers f. 67. Isacoustic Lines. Isacoustic lines are lines which pass through all points at which the percentage of audibility of the earthquake-sound is the same. For their construction, the sound-area is divided into equal areas, and the percentage of audibility within each is supposed * Davison (1), pp. 39-i:}; Hayleigh, Nature, vol. 56, 1897, p. 285. •j- Davison (2), pp. 505-506. 60 SOUXD-PHENOMENA OF EARTHQUAKES [ch. to be equal to that at its centre. Curves corresponding to different percentages are then drawn through the points at Avhich such percentages are affixed, or through points Avhich divide the hues joining successive centres in the proper pro- portion. Fig. 28 shows the isacoustic hues corresponding to percentages 95 and 90 for the Derby earthquake of 1903 (see also Fig. 51). The significance of these lines Avill be referred to in a later section dealing with twin earthquakes (sect. 242)*. Fig. 28. Map of the Derby earthquake of Mar. 24, 1903. 68. Variation in Character throughout the Sound-Area. Throughout the soiuid-area, the sound maintains its uniform lowness of pitch, if we may judge from the frequency with which the word "heavy" is used in descriptions of the sound. As the distance from the epicentre increases, there is a steady decline in the percentage of references to thunder, loads of stones falling and explosions, a steady increase in the references to wind, and on the whole an increase in the frequency of comparisons to passing waggons. Among the comparisons to * Davison (2). pp. 506-508. IV] SOUXD-PHENOMEXA OF EARTHQUAKES 61 thunder, there is a steady and rapid increase in the percentage of references to distant thunder. Thus, there is a greater mono- tony, an approach to iniiformity both in intensity and pitch, as the distance from the epicentre increases*. 69. Relative Magnitude of Sound-Area and Disturbed Area. The relative magnitude! of the sound-area and disturbed area ranges continuously betAveen the widest limits. On the one hand, the shock is felt but is imaccompanied by sound; on the other hand, the sound is heard Avithout any attendant shock. In the great majority of strong and A'iolent earthquakes, the sound-area occupies a region surrounding the epicentre, while the disturbed area extends beyond it in every direction. For instance, in the Verny (Turkestan) earthquake of 1887, the disturbed area contained about 400,000 sq. miles, the soimd- area about 132.000 sq, miles. In the Italian earthquake of 1873, the two areas contained respectively 227,000 and 22,000 sq. miles. In Japan, 30 per cent, of the earthquakes during 1885— 1892 which disturbed areas of more than 10,000 sq. miles were unaccompanied by recorded sound ; and, when heard, the sound as a rule was inaudible at more than a few miles from the epicentre. Of the earthquakes which originated beneath the land during these years, 26-5 per cent, arc recorded as accom- panied by sound. For the submarine earthqiiakes of the same period, the corresponding percentage is 0-84. None of the earth- quakes originated at a greater distance than 40 or 50 miles from the shore, while the epicentres of 93 per cent, of the total lumiber were not more than 10 miles distant. In strong British earthquakes, the disturbed ai'ca extends l)eyond, but not far beyond, the sound-area in all directions. For instance, in the Hereford earthquake of 1896, the disturbed area contained 98,000 sq. miles, and the sound-area 70,000 sq. miles. On an average, the sound-area is about two-thirds of the disturl)cd area. In British earthquakes of moderate strength, and in some slight earthquakes, the sound-area and disturbed area practically coincide. The areas in such cases range from K)0 to 2000 sc|, mik's, and in one case to 43 K) sq, miles. In many slight British earthquakes, the sound-area o\erlaps the disturbed area, as a rule on one side only, sometimes in every direction. Examples of the j)artial overlapping will be given in sect. 72, * Davison (2). |)|). 508-513. 62 SOUND-PHENOMENA OF EARTHQUAKES [ch. Still lower in the scale are earth-sounds, in which sounds alone are observed without the slightest accompanying tremor. As a general rule, they appear to form part of the series of after- shocks of a great earthquake, or occur as intercalated members of a series of weak shocks. For instance, 3365 after-shocks of the Mino-Owari earthquake were recorded at Gifu from Oct. 28, 1891, to the end of 1893, and of these 409 were earth-sounds. After the Comrie (Perthshire) earthquake of Oct. 23, 1839, one observer at Comrie noted, between this date and the end of 1841, 44 shocks and 234 earth-sounds. The earth-sounds in this latter district are of considerable interest. They are heard in an area in which slight shocks, accompanied by precisely similar sounds, are at times very fre- quent. Here there is a complete continuity from earthquake to earth-sound ; from the strong earthquake in which the disturbed area extends in all directions beyond the sound-area, through the moderate earthquake in which both areas coincide approxi- mately, and the slight earthquake in which the sound-area over- laps the disturbed area in one or every direction, down to the earth-sound when the disturbed area vanishes*. 70. Earth-Sounds. In addition to the examples given in the preceding section, reference may be made to three cases in which earth-sounds were especially numerous. In the island of Meleda, in the Adriatic Sea, earth-sounds wei*e frequently heard during the years 1822-1826. Partsch gives a list of the shocks and sounds observed from Nov. 17, 1824, to Feb. 18, 1826. In this interval, there were 30 shocks and 71 detonations. Of the shocks, all but three were accom- panied by sound. Another district in which earth-sounds were at one time fre- quent is that surrounding East Haddam in Connecticut. Before the English settlements, the sounds were well known to the Indian inhabitants, who called the place Morehemoodus or place of noises. According to an observer, writing in 1729, eight or ten sounds, resembling small arms, were sometimes heard in 5 minutes, and great numbers in the course of a year. Often they could be heard "coming down from the north imitating * Davison (1), pp. 53-56; (2), pp. 513-515; F. Oniori, Journ. Coll. Sci., Imp. Univ. Tokyo, vol. 7, 1894, p. 113; J. Drummond, Phil. Mag., vol. 20, 1842, pp. 240-247. IV] SOUND-PHENOMENA OF EARTHQUAKES 63 slow thunder, until the sound came near, or right luider, and then there seemed to be a breaking, like the noise of a cannon- shot, or severe thunder, which shakes the houses and all that is in them."" At the end of the eighteenth century, they were still heard frequently. At the present time, they seem to have ceased. Lastly, Humboldt describes a remarkable series of earth- sounds as the subterranean thunder of Guanaxuato, a city on the Mexican plateau, far removed from any active volcano. From Jan. 13-16, 1784, "it seemed to the inhabitants as if heavy clouds lay beneath their feet, from w hich issued alternate slow rolling sounds and short quick claps of thunder." In this case, the earth-sounds were not accompanied by sensible shocks *. We may infer from these and other examples: (i) that earth- sounds are heard generally in those districts in which slight shocks are frequent; (ii) that, in the midst of a series of earth- sounds, slight shocks, accompanied by precisely similar sounds, are occasionally intercalated, there being a complete continuity from earthquake to earth-sound. We may therefore conclude that earthqviakes and earth-sounds are manifestations, differing only in degree and in the method in which we perceive them, of one and the same phenomenon. 71. Brontides. In many parts of the world, there are heard sounds closely resembling those described in the preceding para- graphs. In the delta of the Ganges they are known as Barisal- guns, on the coasts of Belgium as mist-poeffeurs, in central Italy as marinas, in Haiti as gouffrcs. They have been called brontides by Alippi, who has paid special attention to the subject. The sounds are invariably deep and last as a rule for 5 or 6 seconds; they begin feebly, grow rapidly in strength, and then as rapidly die away. They bear a close resemblance to the booming of guns or distant thimder, though other types of the Davison scale arc referred to. In Italy, for instance, 11 per cent, of the descriptions collected by Alippi are referred to passing waggons, 41 per cent, to thunder, 10 per cent, to wind, 6 per cent. * P. Partsch, lierichl iiber il(i.s IhtoiKitioiis-Plianoun'n (luf tier In.sel Meleda bet/ liagusu (Wien, 1820), pp. 2()J-'211; W. T. liri>rliani. Mem. liosUm Sac. Nal. Hist., vol. 2, 1871, pp. 14—16; A. von Humboldt, Cosmos (Bohn's edition), vol. 1, i)p. 2(>:5, 20.>-20(5. 64 SOUND-PHENOMENA OF EARTHQUAKES [ch. to loads of stones falling, 3 per cent, to the fall of a heavy body, 21 per cent, to explosions, and 8 per cent, to miscellaneons sounds. Thus, as regards type of sound, brontides bear some resemblance to the soimds which accompany slight earthquakes. That the sound of brontides is close to the lower limit of audi- bility is evident from the fact, mentioned by Cancani, that they are heard by some and not by others placed in the same con- ditions. Most frequently, single detonations are heard, but they some- times occur in groups. According to 66 per cent, of the observers consulted by Alippi, the soimd appears to travel through the air, according to 25 per cent, through the ground, and according to 9 per cent, through both air and ground. They are not heard with any approach to uniformity through- out any comitry. Alippi has mapped the districts in Italy that are subject to brontides, and his map shows that they are con- fined to special regions, two or three embracing a large part of a province, the majority small. Certain districts, such as the western Alps, are quite free from brontides. In Italy, brontides are heard with nearly equal frequency in all seasons of the year, though rather more frequently in summer ; in the Philippine Islands, they are most frequent in the hot season (March, April and May). They occur most often about simset and sunrise, least often at night; but this concentration at certain hours may be apparent only, and depending on the habits of the observers. Notwithstanding the attention that has been paid to bron- tides, their cause is still somewhat obscure. Cancani urges that they cannot be due to stormy seas, for they are observed often with a calm sea and at considerable distances from the coast. Nor can they be caused by gusts of wind rushing through movmtain gorges, for they are heard indifferently on the summits of mountains, on the coast and in open plains, and most fre- quently when the air is still. It seems improbable that their origin is connected with the atmosphere, for. in that case, they should be heard everywhere, and not in sjjecial regions only. An artificial origin is excluded, for they are observed at times when guns and mines are not fired, and in places (such as some African countries) where explosives are unknown. IV] SOUND-PHENOMENA OF EARTHQUAKES G5 Cancani thus concludes in favour of a seismic origin for brontides. In support of this, he urges that brontides pre- dominate in countries which are subject to earthquakes, that they are often heard as heralds of earthquakes, and are specially frequent during seismic series, and that brontides are sometimes accompanied by very feeble tremors. Again, in the south-west of Haiti, brontides are very common, especially in the mountain range of La Selle. On the north side, this range is boimded by a steep cliff formed by displacements along a fault that is be- lieved to be still growing. The sounds appear to come from the base of this cliff, and, as they do not differ from those which accompany sensible earthquakes, it is probable that they are caused by small readjustments of the crust along this fault. On the other hand, in the Philippine Islands, according to Saderra Maso, brontides have little apparent connexion with earthquakes. The Belgian coasts are rarely, if ever, visited by local earthquakes. Moreover, Alippi has compared his brontide map of Italy with Baratta's seismic map of the same country, and he finds that, while the areas in which brontides are common are in many cases (such as southern Calabria) the same as those in which earthquakes are numerous and strong, yet that some of the brontide areas are free from earthquakes, while brontides are unknown in some seismic areas. It seems difficult, therefore, to avoid the conclusion that bron- tides have more than one origin, but that in many cases they must be regarded as the same in nature and origin as the earth- sounds described above. It should also be remembered that they are not of necessity the results of recent earth-movements. It is possible that they may be the latest representatives of a series of after-shocks of some long-past and almost forgotten earthquakes*. 72. Relative Position of Sound-Area and Disturbed Area. The cxcentricity of the sound-area with respect to the isoscismal lines is one of the most significant phenomena of earthcjuake- sounds. In the strong earthquakes of this country, the isa- coustic lines show a remarkable independence of the isoseismal lines. It is in the moderate and slight earthquakes, however, * Alipi)! (2), pp. 21-t.l; Cancani (1), pp. 231-234; J. Scherer, Bull. Sets. Soc. Arner., vol. 2, 1912, pp. 230-232. D. M.S. 5 66 SOUND-PHENOMENA OF EARTHQUAKES [ch. that the excentricity is manifested most clearly by the sound- area overlapping an isoseismal line or the boimdary of the dis- turbed area in one direction. For instance, in the Bolton earthquake of 1889, the isoseismal lines (indicated by the continuous lines in Fig. 29) are nearly circular, the boundary of the sound-area (indicated by the dotted Scale of Miles 6 ~' ' Fig. 29. Map of the Bolton earthquake of Feb. 10, 1889. line) is also nearly circular, but its centre lies 3 J miles south- south-west of the centre of the isoseismal 5. The great Irwell Valley fault, the course of which is represented on the map by the broken line, hades to the north-east and therefore beneath the epicentre. As the earthquake was probably caused by a slip along this fault in the neighbourhood of Bolton, it follows that the sound-area with respect to the isoseismal 5 is shifted towards the fault-line. IV SOUND-PHENOMENA OF EARTHQUAKES 67 Again, in Fig. 30, the continuous lines represent the isoseismals 3 and 4 of the Helston earthquake of 1898, and these show that the originating fault must hade to the south-east (sect. 129). The outer dotted line indicates the boundary of the sound-area ; and the inner dotted line, which is concentric \Wth the other, separates the places where the sound was very loud from those where it was distinctly fainter. In this case. also, the sound-area S:ala or yile I 1 1 1 1 12 3 4 Fig. 80. Map of the Helston earthquake of Apr. 1, 1898. relatively to the disturbed area is displaced towards the fault- line*. 73. Time-Relations of the Sound and Shock. The earthquake- sound almost invariably accompanies the shock. In British earthquakes, it usually also i^recedes and follows the shock, overlapping it by one or a few seconds at both ends. This over- lapping persists in all parts of the sound-area, for, in the four zones bounded by successive isoseismals, the beginning of the * C. Davison, Geol. Mag., 1891, pp. ;J0G-;J1(>; Quart. Journ. Geol. Soc, vol. 50, 190(), pp. 1-7. 5—2 68 SOUND-PHENOMENA OF EARTHQUAKES [ch. iv sound precedes that of the shock in 67, 69, 70 and 62 per cent. of the records respectively; while the end of the sound follows that of the shock in 38, 44, 41 and 36 per cent, in the same zones. It would seem, then, that the precedence of the sound is due to a difference in the place of origin rather than to the sound-waves travelling with a greater velocity than the large waves *. * Davison (2), pp. 517-522. CHAPTER V DEFORMATIONS OF THE EARTH'S CRUST 74. The deformations of the crust observed with some great earthquakes take the form of (i) fault-displacements, and (ii) Avarping of the surface-beds. They are thus of the same character as those which occur in lower layers of the crust, but, in the more rigid outer crust, faulting, as might be expected, predominates over warping*. The connexion between the crustal deformations and the earthquakes is shown by the coincidence between their times of occurrence and the areas affected by them. It is important, however, to notice that the deformations are not consequences of the earthquakes, but rather, as will be seen in Chapter XIV, primary causes of the earthquakes. The number of earthquakes accompanied by crustal deforma- tions is considerable. In many cases, however, while it is needless to doubt the reality of the permanent movements, the observa- * The more important memoirs in which deformations of the eartli's erust are described are the followino: 1. Fuller. M. L. The New Madrid eartluiuake. Bull. U.S. Geol. Surv., No. 404. 1012, pj). 1-112. 2. Hobbs, W. H. The earthquake of 1872 in the Owens Valley, Cali- fornia. Beilr. zur Geopfi., vol. 10, 1910, pp. 3.52-385 (especially pp. 371-384). 3. Koto, B. The cause of the fjreat earthquake in Central Japan, 1891. Jimrn. Coll. Sci., Imj). I'niv. Tokyo, vol. 5, 1893, i)p. 295-353. 4. Lawson, A. C. (editor). The Califi>riii(iu Kiirllniudkc of .April 18, 190G, vol. 1 and atlas, 1908; vol. 2 (by II. V. Heid). 1!)1(). .-). Lyell, C. Principles of Geolo3-(i9 (see also pj). 70-72). 8. Tarr, H. .S., and L. Martin. The eartlupiakes at Yakutat Hay, .Alaska, in Se|)tember 1899. U. S. Geol. Surv., Prof. Paper No. 69, 1912, pp. 1-135 (esjiecially p|). 18-45). 9. ReUizione delUi Cotnniinsione lieole ineurieala di desifinare le zone piii adatti jtcr la rceonslruzioue deffli abilati eol/iili dal terremolo del 28 dicevibre 1908, ecc. Homa, 1909, pp. 131-156. 70 DEFORMATIONS OF THE EARTH'S CRUST [ch. tions recorded add little to our knowledge beyond the fact that some movement, usually of elevation, has taken place. Elimi- nating all such earthquakes, there remain fourteen in which the accompanying displacements have been observed and described with some care. These earthquakes are those of: (1) New Madrid (U.S.A.) in 1811-12; (2) Wellington (N.Z.) in 1855; (3) Owens Valley (U.S.A.) in 1872; (4) Sonora (U.S.A.) in 1887; (5) Mino- Owari (Japan) in 1891; (6) Sumatra in 1892; (7) Baluchistan in 1892; (8) Locris (N.E. Greece) in 1894; (9) Assam in 1897; (10) Alaska in 1899; (11) Kangra (India) in 1905; (12) Formosa in 1906; (13) California in 1906; and (14) Messina in 1908. The displacements of the first and third of these earthquakes left traces which are still visible; in the other cases, they were observed and measured as a rule soon after their formation. Fault-Displacements 75. The principal features of the fault-movements which occvu" during earthquakes are the great length of the fault over which the movement takes place and the general uniformity in the direction of the fault-line. The displacement may be almost ' entirely horizontal or almost entirely vertical, but in most cases both horizontal and vertical. Horizontal displacements are usually manifested by the relative shifting of objects previously in contact or in line; vertical displacements by the formation of fault-scarps. When the displacements are of considerable magnitude, a new trigonometrical survey of the central district, if one has been made not many years before, may convert relative into absolute measurements. One other feature may be referred to at this stage, namely, that the faulting occurs repeatedly along the same line. The Baluchistan fault has the appearance of an old road, and the natives declare that the ground always cracks along the old fault-line with every severe earthquake. The Owens Valley con- tains many scarps formed before 1872, the height of one of which was doubled in that year. The southern part of the San Andreas fault in California lies in the desert part of the Coast Ranges, in which erosion takes place slowly. With every great earthquake, the fissure opens anew; so that, to the inhabitants of the district, the fault is known as the "earthquake-crack." V] DEFORMATIONS OF THE EARTH'S CRUST 71 Of the fourteen earthquakes referred to in sect. 74, the de- formation in all but three (those of New Madrid, Kangra and Messina) took the form of faulting. In the exceptional cases, the only phenomenon observed was that of warping, either local or general, but it is not impossible that this warping was merely the surface equivalent of deep-seated faulting. In two other Fip. ;J1. Fault -sea i|) of tlu' .Miiio-Owari cartlKiuake of 18J)]. earthquakes (Assam and Alaska), both faulting and warj^ing were present, and this seems to lend some support to the con- nexion. 76. General Appearance. The superficial effects of the faulting vary with the direction of the movement and also with the nature of the ground traversed. In the case of the Sumatra 72 DEFORMATIONS OF THE EARTH'S CRUST [ch. N.10°E, earthquake of 1892, there was no actual trace of the fault at the surface, and its existence was only revealed by geodetic measurements. In all the other earthquakes, one or more of the faults concerned could be followed for some distance. When the displacement is •^ "^^ ^ horizontal or nearly so, the fault I apjDcars as a fine crack in hard rock, as in parts of the Mino- Owari fault, or as an open fissure in earthy ground, as in the Locris fault. In such cases, the relative shifting of objects on either side of the fault is the clearest evidence of the displace- ment. When the displacement is partly or wholly vertical, the superficial effects are more dis- tinct. If the surface consist of soft earth and if the uplift be small, the fault appears like a rounded ridge, from 5 to 10 feet wide and about 2 feet in height, as if the soil had been raised by a gigantic mole creeping luider- ground (Fig. 31). In the Cali- fornian earthquake of 1906, the formation of this ridge was due partly to the shearing action and partly to compression along the line of fault, and was in many places accompanied by a series of secondary cracks extending a few hundred feet from the fault. Some of these are shown in Fig. 32, the broken lines representing the directions of the secondary cracks which are inclined to the direction of the fault-line at an angle of 42°. Uplifts of more than 2 feet in amoimt result in the formation of fault-scarps. In compact rock, the scarps appear as long cliffs, which in the cases known, are always vertical (Fig. 33). In alluvial or earthy ground, the scarps soon weather down into Fig. 32. Secondary cracks in the Californian eartliquake of 1906. V] DEFORMATIONS OF THE EARTH'S CRUST 73 Fig. 33. Minui fault, Alaskan eaitliquake, 1899. Viii. '.\l. l""aiilt-s(ar|) at .Midori of the .Miiio-Owari tai tluiuakc of 1891. 74 DEFORMATIONS OF THE EARTH'S CRUST [ch. a uniform slope, so that from a distance they resemble railway- embankments (Fig. 34). If, however, the alluvium be of great thickness, the displacement may fail to reach the surface, and the scarp is then replaced by a slope, that is, by warping. For instance, the Chedrang favilt, formed during the Assam earth- quake of 1897, runs at its north end (Fig. 38) beneath a thick mass of alluviimi, converting what was level ground into a smooth unbroken slope, on which trees are tilted over, as shown in Fig. 35. 77. Length. The length of the fault-displacement varies be- tween wide limits. It is least when the fault-system is complex, greatest when the displacement occurs along a single fault. Fig. 35. Slope in alluvium over the Chedrang fault. Thus, in the Assam earthquake of 1897, there Avere two prin- cipal faults, the Chedrang fault not less than 12 miles in length, and the Samin fault about 2| miles long. In the Alaskan earth- quakes of 1899, the longer faults {A, E and B, Fig. 39), inferred from the evidence of the vertical displacements, are 18, 16 and 13 miles in length resiDectively, though it is possible that the first and third form a single fault 31 miles long. In the following earthquakes, the movement at the surface, was practically confined to a single fault in each case; though possibly, as in the Mino-Owari earthquake of 1891, more than one deep-seated fault may have been in action. In the Balu- chistan earthquake of 1892, the movement affected a distance v] DEFORMATIONS OF THE EARTH'S CRUST 75 of several miles, and the fault itself has been traced for 120 miles. The Formosa fault (1906) was also traced for only part of its length; the entire displacement is, however, estimated by Omori at about 30 miles. The Locris fault (1894) was 34 miles long, the total length of the Owens Valley fault-system (1872) about 40 miles. The Mino-Owari fault (1891) was traced for 40 miles, but there can be little doubt that its length at the surface was 70 miles. The Wellington fault-scarp (1855) was a mainly vertical cliff 90 miles in length. The fault connected with the Sumatra earthquake of 1892 was detected only by the re-triangulation of the district; its length was certainly much greater than 34 miles, and, according to Reid, may have been as much as 125 miles. Xo more remarkable displacement is known to haxe occurred than that which accompanied the Californian earth- quake of 1906. Though lying for about 70 miles under the sea, the total length was almost certainly 290 miles*. The displacement during any given earthquake does not ne- cessarily occur over the whole length of the fault concerned. On this point, our information is limited to the faults of the Owens Valley and Californian earthquakes, the latter of which has been studied in greater detail than any other. This, the San Andreas fault, has been traced, with three interruptions in which its course is submarine, from near Cape Mendocino on the north, past San Francisco, to the north end of the Colorado Desert on the south, that is, for a distance of more than 600 miles. During the earthquake of 1906, the movement was confined to its nf)rthern half, from near Cape Mendocino to San Juan, 82 miles south-cast of San Francisco f. 78. Form. The form of the faults with Avhich earthquakes are connected is similar to that of the faults determined by other evidence. Small faults, such as the Chedrang and Samin faults of the Assam earthquake, are practically straight. Longer faults, such as those in Baluchistan and Locris, maintain a nearly uniform direction through their entire known courses. The * Oldham, pp. 146-148; Tarr and Martin, p. 35 and plate 14; C. A. and A. II. McMahon, Quart. Jouni. Gcol. Sue, vol. 53, 1H97, p. 2<)2; Omori, pp. 58-Gl ; S. A. Papavasiiiou, Compl. Rend. Acad. Sci. Paris, vol. 11<), 1894, p. 114; Ilobbs, pp. 378-381; Koto, p. 349; Lyell, p. 86; Heid, pp. 76-77; Lawson, vol. 1, p. 54. f I^awson, vol. 1. pp. 48, 54. 76 DEFORMATIONS OF THE EARTH'S CRUST [ch. C.Mendocino Pb.Delgada direction of the longest of all, the San Andreas fault, between San Juan and Point Arena varies only from 30° to 40° W. of N. (Fig. 36). When drawn on a map of small scale, such as that of Fig. 36, the fault as a whole appears as a nearly even line, slightly curved and convex to the Pacific. On a large-scale map, however, it is seen that the fault is not a smooth uniform curve, but a succession of slightly curved rather than straight por- tions, the curvatures varying in direction. In a few cases, the faults are not single. Though only one great scarp was observed with the Mino-Owari earthquake of 1891, there must have been a second and more deeply-seated fault along which most of the after- shocks originated, but of which no trace appeared at the surface. The main fault of the Formosa earthquake of 1906 ended to- wards the west in a branch fault (Fig. 40). The Owens Valley earthquake of 1872 was con- nected with a system of faults. For one or two miles individual scarps maintain a nearly constant direction, but in some parts they are subject to abrupt changes of direction, so that their course is a succession of zigzags with sharp elbows; in others, they are arranged in parallel lines, slightly overlapping*. 79. Relation to the Form and Structure of the Ground. When viewed in detail, the course of an earthquake-fault seems to * C. Davison, Beitr. zur Geoph., vol. 12, 1912, p. 10; Omori, p. 57; Hobbs, pp. 374-383. San Francisco Fig. 36. Map of the San Andreas fault. v] DEFORMATIONS OF THE EARTH'S CRUST 77 be independent of the form of the ground; and, in this respect, it differs totally from the earth-fissures described in Chapter VII. After running for considerable distances along a valley, the Mino-Owari fault crosses hill-spurs and in one case the top of a hill. Throughout its entire length, the San Andreas fault in California lies along depressions or at the base of steep slopes, which are due partly to erosion and partly to displacement along the fault. But its position with regard to the mountain-ridge frequently ^•aries, as it passes several times through breaks in the chains from one flank to the other. On a large scale, however, the course is dependent on the structure of the district. The Baluchistan earthquake-fault, for instance, runs in a nearly straight line parallel to the Khojak Range. The Locris earthquake-fault follows a nearly constant east-south-east direction parallel to the Gulf of Euboea. The Wellington earthquake-fault is represented by a continuous escarpment running along the foot of the Remutaka Mountains, where they present a steep slope towards the great Tertiary plain of the Wairarapa. One of the Alaskan earthquake-faults runs along the main portion of Russell Fiord (Fig. 39), another at the foot of the straight mountain front on the east side of Yakutat Bay*. 80. Horizontal Displacement. The horizontal displacement is usually shown by the dislocation of fences, roads, bridges, tunnels, pipes or any structure that crosses the line of the fault. Two examples of this displacement are illustrated in Figs. 34 and 37. In the former, the upper part of the severed road is shifted 13 feet to the left by the Mino-Owari fault; in the latter, the two portions of the fence which lie on opposite sides of the San Andreas fault are now scj:)arated by se\'eral feet. Without a duplicated trigonometrical survey, it is usually impossible to determine which side has moved or whether both sides have moved. The relative displacement of the two sides is, however, so far as known, constant indirection throughout agiven fault. In the Locris earthquake, the north-east side was shifted to the north-west relatively to the other side; and this was also the * Koto, pp. :5:{:J, '.V.W, 3:^8, :J41, :J41-; Luwson, vol. 1, pp. 48-;V2; C. Davison. Geol. Mag., 1803, p. :55J>; S. A. Papavasiliou, Compt. Rend. Acad. Sci. Paris, vol. 119, 1894, p. 380; Lyell, p. 80. 78 DEFORMATIONS OF THE EARTH'S CRUST [ch. case in the Mino-Owari earthquake. In the Bakichistan earth- quake, the east side moved towards the north. In the Formosa earthquake, the north side was shifted relatively eastwards. Throughout the whole vast extent of the San Andreas fault, the displacement of the north-east side in 1906 was to the south-east. The amounts of the horizontal displacement are sometimes considerable. In the Baluchistan earthquake, it was not less than 2^ feet; in the Formosa earthquake it varied from 2 to 8 feet; in the Mino-Owari earthquake from 3 to 13 feet; in the Fig. 37. Fence severed by the San Andreas fault-displacement in the Californian earthquake of 1906. Owens Valley earthquake from 3 to 12 feet, and in places to 18 and 20 feet; in the Californian earthquake it lay as a rule between 8 and 15 feet, but in one place amounted to 21 feet. In a few cases, the shifting parallel to the fault was accom- panied by actual compression of the ground in the perpendicular direction. There is some slight evidence of this in the Baluchistan earthquake, and distinct proof of it in the Mino-Owari earth- quake. Plots of ground that were 48 feet in length before the earthquake were reduced to 30 feet afterwards. Indeed, according V] DEFORMATIONS OF THE EARTH'S CRUST 79 to Milne, it would seem that the whole Xeo Valley had become narrower*. 81. Vertical Displacement. The displacement of the ground in a vertical direction is usually conspicuous in the form of fault-scarps (Figs. 33, 34). Here, again, the measured displace- ment as a rule is relative only. A revision of the levels in a direction across the fault might determine which side, or whether both sides, had mo\ed. In default of such precise evidence, the effects of the movement in changing the gradient of streams may be of service, as in the Mino-Owari and Assam earthquakes. In one case only, that of the Alaskan earthquakes of 1899, are the absolute changes of elevation known. The epiccntral district being intersected by Yakutat Bay and its branches, the uplift could be measured even six years after the earthquakes by the heights of dead barnacles and mussels still clinging to the cliffs, . and the subsidence by that of the base of the lowest dead tree in place. The measured vertical displacements, whether relative or absolute, are in some cases less, in others far greater, than the horizontal movements. In the Califomian earthquake the greatest uplift was 3 feet, in the Formosa earthquake 6 feet, and in the Wellington earthquake about 9 feet. In the Mino-Owari earthquake, the greatest uplift was about 10 feet, except at Midori (Fig. 34) where it amounted to nearly 20 feet. In the Owens Valley earthquake, one uplift of 23 feet was measured. The Chedrang fault-scarp of the Assam earthquake attained heights of 32 and 35, feet. The greatest known uplift is that of 47 ft. 4 ins. on the north-west shore of Yakutat Bay in Alaska. In any one fault-scarp, the uplift may vary greatly. For in- stance, the length of the Chedrang fault mapped in Fig. 38 is about 12 miles, but, even in this short distance, there are three undulations, separated by intervals of no displacement, in which the maximum uplifts are 25, 35 and 32 feet, respectively. In a few cases, there may even be reversal in the direction of throw, as in the Mino-Owari fault-scarp at Midori (Fig. 34), where the north-east side was ele\ated nearly 20 feet, whereas, in all other * C. Davison, GvnL Mfig., 189:3, p. :J.>5); Omori, pp. .jT-.jit; Koto, pp. :{;$0- :j4(i ; Hobbs, pp. :$79-:J80; Lawson, vol. 1, pp. 52-80; .J. Milne, Seis. Jutini., vol. 1, 1893, p. 131. 80 DEFORMATIONS OF THE EARTH'S CRUST [ch. parts of the known scarp, the uj^hft occurred on the south-west side *. 82. Surface Effects of Faulting. In the neighbourhood of fault-scarps, the ground is intensely fissiu-ed and landslips are almost con- tinuous. But the most marked effects of faulting (and, to a less extent, of warping) are those on the flow of streams. If the uplift take place across a river-bed, the result is a decrease of gradient for a short distance upstream and an increase similarly downstream. The former effect is the more impor- tant, the river being often widened out or ponded back by the obstruction into a small lake or pool. Such effects are common to all cases of faulting, but were unusually con- spicuous along the course of the Chedrang fault formed during the Assam earthquake of 1897 f. The course of this fault, as will be seen by the broken-line in Fig. 38, is jDractically straight for a distance of at least 12 miles. The figin-es on the right-hand side indicate the throw in feet in different parts of its course, * Lawson, vol. 1, pp. 80-87, 140-145; Omori, pp. 57-59; Lyell, pp. 85- 86; Koto, pp. 330-346; Hobbs, pp. 378-379; Oldham, pp. 138-148; Tarr and Martin, pp. 18-45. t A photograph of one of these pools is reproduced in Marr's Scientific Study of Scenery, 1900, plate facing p. 180. Fig. 38. Map of the Chedrang fault. v] DEFORMATIONS OF THE EARTH'S CRUST 81 the rock on the east side, except where there is no displacement, being invariably the higher. Throughout its known course, the fault runs along the valley of the Chedrang river, which flows from south to north. Thus, whenever the stream crosses the fault-line from east to west there is a waterfall, as at A, etc. When the stream meets the fault-scarp from the west, it is ponded back and forms pools, as at B, etc. Pools also collect on the west side of the fault when small tributary streams meet the scarp, as at C, etc. Besides these pools, there are others, at D, D, with a different origin. Each is about half a mile long, there is no visible barrier, and in one case the pool spreads across the fault. In both, the channel of the stream sinks in the upstream direction, without trace of faulting, beneath the waters of the pool. Now, it is just where the fault has no throw that the pools are widest and deepest. The pools must therefore be due to warping or the formation of an undidation across the stream-course, by which the natural slope of the ground has been reversed*. Natire of Fault-Deformation 83. The nature of the fault-deformation is definitely known in at least eleven earthquakes, but, even in this small number, it varies considerably. There seem to be at least four kinds of deformation, (i) In the first class, the movement is almost entirely horizontal, or the horizontal displacement is greatly in excess of the vertical. To this class belong the Californian, Sumatran, Baluchistan, and, possibly, the Locris, earthquakes, (ii) In the second class, the vertical displacement predominates or occurs without any horizontal shifting. The Wellington, Assam, Alaskan, and possibly the Owens Valley, earthquakes are typical of this class, (iii) The third class includes the Mino- Owari and, |M)ssil)ly, the Sonora, earthquakes, in which both horizontal and ^■crtieal movements occur to approximately the same extent, (iv) The fourth class includes, so far as is known, only the Formosa eartlujuake and is one of great interest, the vertical displacement being in opposite directions in the two halves of the fault. * Ol.lli.im. |)i). m.S-M-S. 82 DEFORMATIONS OF THE EARTH'S CRUST [ch. 84. Movements chiefly Horizontal. In the Californian earth- quake, the vertical displacement was small compared with that in a horizontal direction, and to the south of San Francisco was imperceptible. No vertical displacement was detected with the Sumatra earthquake. In the Baluchistan earthquake, the only measurement of uplift amounted to about 2 inches. The question as to which side moved or whether both sides moved is one that can only be answered by a renewal of the trigonometrical survey of the district, and this was accomplished for the Californian earthquake before the lapse of a year. The area covered by the survey is about 170 miles long and 50 miles in maximum width. It extends from a line a short distance to the south of San Juan to the neighbourhood of Fort Ross (Fig. 36). It was found that since 1851, when the first survey was begun, the stations in the neighbourhood of the fault had all been displaced by amounts ranging from less than a foot to 19| feet. Though more than half a century had elapsed between the dates of the two surveys, there can be little doubt that the movements revealed by the re-triangulation occurred in 1906. (i) The horizontal displacements at the various stations were nearly always in directions parallel to the fault, (ii) The total dis- placements of stations close to the fault were approximately equal to those observed in lines of road or fencing severed by the fault. The most important points established by the new survey are: (i) that both sides of the fault were displaced, the south- west side to the north-west, and the north-east side to the south-east; and (ii) that the movements were not confined to the immediate neighbourhood of the fault, the displacement de- creasing on both sides with increasing distance from the fault. For instance, on the north-east side of the fault, ten points at an average distance of a little less than a mile from the fault have an average displacement of 5-1 feet to the south-east; three points at an average distance of 2| miles have moved on an average 2-8 feet in the same direction; while one point at a distance of 4 miles has a displacement of 1-9 feet. No point on this side at a greater distance than 4 miles has suffered any displacement distinctly exceeding that due to errors of observa- tion. On the south-west side of the fault, twelve points at an V] DEFORMATIONS OF THE EARTH'S CRUST 83 average distance of 1;|: miles have an average displacement of 9-7 feet to the north-west; seven at an average distance of 3| miles have one of 7-8 feet; while one point 23 miles from the fault was displaced 5-8 feet. Thus, straight lines drawn on the surface on either side of the fault and at right angles to it would after the earthquake become slightly curved, the concavity facing the south on the north-east side, and facing the north on the south-west side. Moreover, for points on opposite sides of the fault and at equal distances from it, the displacements on the south-west side were t^ice as great as those on the north-east side up to a few miles from the fault. The first earthquake in which sudden displacements were afterwards established by geodetic measurements was the Su- matra earthquake of 1892. The movements were then entirely horizontal. There was no trace of any visible fault at the surface, but the measured displacements, according to H. F. Reid. show that the fault must be directed to the north-north-west, that the crust on the west side was shifted to the north and that on the east side towards the south; that the total relative displacement of the two sides was 11| or 13 feet; and that the displacement diminished rapidly with increasing distance from the fault*. The results of this section may thus be summed up as follows. So far as at present known, (i) the movement when chiefly hori- zontal is confined to strike-faults; (ii) both sides move in opposite directions; (iii) the amount of the displacement diminishes ra]iidly with increasing distance from the fault. 85. Movements chiefly Vertical. In the Assam earthquake, the movements along the two principal faults were entirely vertical. In the Alaskan earthquake, the vertical movements predominated, though there may in parts have been small hori- zontal displacements. An uplift only is recorded during the Wellington earthquake of 1855; but, in this early case, the evidence does not preclude the occiu-rence of small horizontal shifts. Horizontal movements certainly took place during the Owens Valley earthquake, but there is no evidence to show that they were more than local. Of these four earthquakes, the Wellington earthquake was * Liiwsoii, vol. 1, |)|i. 114— 14.j; Held, pp. 72-78. 6—2 84 DEFORMATIONS OF THE EARTH'S CRUST [ch. marked by the simplest type of displacement. On the west coast of North Island, there was no perceptible uplift at a point 16 miles to the north of Wellington. To the sonth of this point and eastwards along Cook Strait, the amoimt of upheaval increased gradually imtil it reached 9 feet on the east side of Port Nicholson and along the east flank of the Remutaka Moun- tains, where it formed a fault-scarp 90 miles in length. At the Fig. 39. INIap of the Yakutat Bay earthquake-faults in 1899. southern end of this scarp, the uplift of 9 feet was measiu-ed with reference to the level of the sea by the elevation of a white band of nullipores ; and the same evidence showed that, to the south at any rate, the Wairarapa plain on the east side of the scarp was undistiu'bed. The effects of the movements were thus to uplift the Remutaka Moimtains by about 9 feet, and to tilt a tract of country, 90 miles long and 23 miles wide, slightly to the westward. V] DEFORMATIONS OF THE EARTH'S CRUST 85 The displacements of the Alaskan earthquake were of a more complex character. Along the coast, they were probably con- fined to Yakutat Bay and its branches (Fig. 39). The total length of coast-line of this inlet is about 150 miles. For a third of this distance, there was either no change or a very small change of level. These portions are indicated in Fig. 39 by small circles. Elevated regions, by far the more numerous, are denoted by figures only, the depressed regions by figures with minus signs. The figures show the amount of the uplift or subsidence in feet and inches. It will be noticed how rapidly the amount of the uplift varies within a short length of coast. For instance, on the west coast of Disenchantment Bay, the uphft at one point is 42 feet, about a mile to the west 30 feet, and a quarter of a mile farther on only 9 feet. Though no great fault-scarps were noticed, there can be little doubt that these rapid variations were due to faulting rather than to warping, in part perhaps to a series of minor faults like that shown in Fig. 33. The broken-lines in Fig. 39 represent the probable courses of the faults which, according to Tarr and Martin, are implied by the variations in the changes of level. These faults divide the crust into at least three distinct blocks, the known sides of which are roughly parallel. One of these blocks is bounded on three sides by the faults. A, B, C and E, a second by the faults C and G, while the third is bounded by the fault E and includes the north-east shore of the main portion of Russell Fiord. All three blocks extend to unknown distances in other directions than those mentioned. The principal effects of the widespread movements were an uplift of all the blocks along the lines of fault and a slight tilting of the masses away from the faults. The uplift was, however, accompanied by other movements — by a slight depression on the west side of the fault or faults A, B, and in many places by small slips along series of minor faults, due apparently to local adjustments in the tilted blocks*. The results of this section may be summarised as follows: (i) the movements when chiefly vertical arc confined to vertical faults or "blatts"; (ii) either both sides move in opposite directions, or one (the mountainous) side alone is nio\ cd and * Lyell, pp. 82-89; Tarr and Martin, pj). 18-45. 86 DEFORMATIONS OF THE EARTH'S CRUST [ch. uplifted; (iii) the effect of the uphft is to tilt one or more large crust-blocks in the direction away from the fault or favilts. 86. Vertical and Horizontal Movements. The Mino-Owari earthquake was associated Avith a transverse fault which crossed almost the whole width of the Main Island in a general north- west and south-east direction. The relative displacement of the north-east side of the fault in the horizontal direction was in- variably to the north-west and usually ranged from 3 to 13 feet, and in the vertical direction downwards by amounts varying from to 10 feet. In the neighbourhood of Midori, however, or near the centre of the fault, the throw of the fault was reversed, the north-east side beins" raised about 20 feet above the other*. Sharonshl O Fig. 4.0. Map of the Formosa earthquake-faults in 1906. 87. Complex Displacements. The only evidence with regard to such dis})Iacements is that furnished by the Formosa earth- quake of 1906. The plan of the fault, so far as traced, is shown in Fig, 40. The length of the main portion is about 7 miles and of the branch about 2 J miles; but Omori thinks it probable that the branch-fault extended about 7| miles farther west, and that the form of the isoseismal lines indicates an extension of the main fault at least 12 or 16 miles farther east in a moun- tainous tract, so that the total length of the fault would be about 30 miles. Throughout the whole observed length of the fault, the north side was displaced relatively to the east by amounts ranging from 2 to 8 feet; and, throughout the greater * Koto, pp. 330-346. v] DEFORMATIONS OF THE EARTH'S CRUST 87 part (indicated by the short perpendicular hnes), the north side Avas depressed relatively by 3 or 4 feet. Towards the east end, however, the south side was depressed with reference to the other by as much as 6 feet. Assuming this condition to be con- tinued throughout the inferred continuation of the fault, the complex displacement may be represented by Fig. 41, in which the shaded areas denote the portions depressed with reference to the adjoining portions on the other side of the fault, while the arrows indicate the directions of relative shear. The signifi- cance of this form of displacement will be seen when we consider the phenomena of twin earthquakes (sects. 242-244)*. Fio;. 41 . Diagram illustrating the displacements in the Formosa earthquake of 190(i. Warping 88. In warping, the variation in the amount of displacement is contiiuious; there is no sudden break along a definite line, as there is in faulting. The warping may be (i) general, or ex- tending over a considerable area, as in the Kangra earthquake of 1905 and the Messina earthquake of 1908, or (ii) local, as in the New Madrid earthquakes of 1811-1812 and (in part) in the Assam earthquake of 1897. 89. General Warping. The Kangra earthquake of 1905 origi- nated in two foci (Fig. 52), one beneath Kangra and Uharmsala, the other beneath Dehra Dun and Mussoorie, the intensity of the shock at the former towns being far greater than at the latter. There was, however, no visible sign of disturbance at the surface, and, from this and from the imiform distribution of damage in the central areas, it may be inferred (sect. 139) that l)oth the foci were deep-seated. The only line of levels made before the earthquake was that of 1862 from Saharanpur (on the plain) through Dehra Dun to Mussoorie. The jiortion from Dehra Dun to Mussoorie was repeated in 1904, less than a year before the earthquake, and again about a month after it, and * Oinori, |)|). .)7-.>0, 7()-72. 88 DEFORMATIONS OF THE EARTH'S CRUST [ch. these observations indicated that, in the interval, the height of Dehra Dvm with respect to Miissoorie had increased by 5 inches. A fresh series of levels, carried out along the whole line from Saharanpur in 1906-1907 showed that, regarding the height of Saharanpur as fixed, Dehra Dun had risen about 5 inches, while the height of Mussoorie was almost luichanged. Though the amount is a small one, it seems reasonable to conclude that the only superficial effect of the deep-seated movement within the Dehra Dun focus was this very slight buckling-up of the crust. The Messina earthquake of 1908 had also two foci, both lying beneath the Straits of Messina. A line of levellings had been carried round the Calabrian coast during the years 1906-1908, and this was repeated about two or three months after the earthquake from Giaio Tauro, which lies about 20 miles north- east of San Giovanni at the northern entrance to the straits, to Porto Salvo on the south coast. Assuming the height of the former place to have been unaltered, the measurements indicate a subsidence of the whole east coast of the straits, of 17 inches at San Giovanni, 21 inches at Reggio, 24 inches 2 miles south of Reggio, and 14 inches at Pellaro at the southern entrance to the straits. On the ojDposite coast, the subsidence was slightly greater, amounting to 28 inches at Messina*. 90. Local Warping. During the New Madrid earthquakes of 1811-1812, local warping took place on an extensive scale. Ac- cording to Fuller, the earthquakes were caused by displacements along a fault lying about 15 miles west of the river Mississippi, directed about N. 30° E., and about 75 miles in length. In the central district of the earthquake, there are three (if not four) linear troughs each still occupied, though not continuously, by "sunklands," consisting of river-swamps and lakes. The greatest depths of the hollows in which these lakes are situated are from 15 to 20 feet. The troughs are separated by two (if not three) discontinuous ridges, trending in nearly the same direction as the inferred fault. On one of these ridges lie three low flat domes, the largest of which is about 15 miles long and from 5 to 8 miles wide, the surface having been uplifted from 15 to 20 feet. The total length of this ridge is not less than 70 miles. * C. S. Middlemiss, Mem. Geol. Surv. India, vol. 38, 1910, pp. 348-349; Rd. della Com. Reale, pp. 131-156. V] DEFORMATIONS OF THE EARTHS CRUST 89 It should be noticed that, in this case, the observed warping is that of the alluvial beds of the Mississippi plain, and not of the solid crust below. But warping on so large a scale and in so uniform a direction can hardly exist in the surface layer alone. It implies deformation of the crust below, though whether that deformation be in the form of warping or faulting or both is of course unknown. Some interesting examples of local warping occurred in con- nexion with the Assam earthquake of 1897. At several places 'svithin its central area, the form of the hills was perceptibly changed. In other parts, surface undulations altered the gradient of streams, and gave rise to pools or small lakes. Two such pools along the Chedrang fault have been described in sect. 82 (Fig. 38, Z), D). Other groups of pools were formed at a distance from any visible fault, some of them from a quarter of a mile to a mile in length, and one of them more than 20 feet deep. The epicentral area of the Assam earthquake was of great magnitude — according to Oldham, about 200 miles long from east to west and not less than 50 miles in Avidth (Fig. 47). Near the boundary, the displacements were probably long low rolls, resulting in changes in the aspect of the hills; these were suc- ceeded by more pronounced undulations sufficient to reverse the drainage of rapidly-flowing streams ; and these again merged in the central regions into fractures and faults*. 91. Probable Connexion between Warping and Faulting. The distribution of surface warping and faulting in the Assam earth- quake suggests that there may be some connexion between the two phenomena. Fractures, which at the surface show no dis- placement, may have a considerable throw a mile or two below it; and the undulations that gave rise to pools are probably mere folds in the upper rocks produced by faulting below. The slope illustrated in Fig. 35, for instance, is merely an undulation in a thick bed of alluvium due to the passage of the Chedrang fault below. Thus, when faulting takes place at some depth, it is represented either by no displacement or by warping of the outer crust. It is only when the fault-displacement occurs at a small depth that scarps arc left projecting at the surface. * Fuller, pp. 15, 62-75, 105-109; Oldham, pp. 138-163. CHAPTER VI SEISMIC SEA-WAVES 92. Under the general term seismic sea-waves {tsunamis of the Japanese seismologists) are included waves of three kinds: (i) Conde'isational waves, usually known as seaquakes; (ii) Gra- vitational waves, which, under certain conditions, are pro- pagated in all directions from a submarine epicentre with a velocity depending on the depth of the sea; (iii) Stationary waves, the periods of which in different bays are governed by the forms and dimensions of the bays. These are also called marine seiches*. Of these three forms of sea-wave, the first occur with many earthquakes, whether their origin be inland or submarine, and in lakes and rivers as well as in the sea; the second only with great earthquakes of submarine origin; while the third may be excited in bays by the preceding class of waves, but frequentlj^ occur at other times from the action of storms, changes of barometric pressure, etc. The first class of waves or seaquakes consist of condensa- tional vibrations only, water being incapable of transmitting distortional vibrations. In earthquakes of moderate strength, the surface of the water in pools or the sea is ruffled. In strong * The principal memoirs on seismic sea-waves are the following: 1. Geinitz, F. E. Das Erdbeben von Iquique am 9. Mai 1877, etc. Ksl. Leop. -Carol. -Deutschen Akad. der Nahirf. (Halle), 1878, pp. 385-444. 2. Hochstetter, F. von. tJber das Erdbeben in Peru am 13. August 1868, etc. SUzungkh. Akad. Wien,wo\. 58, 1868, pp. 837-860; vol. 59, 1869, pp. 109-132; vol. 60, 1870, pp. 818-823. 3. Honda, K., Terada, T., Yoshida, Y., and Isitani, D. Secondary undu- lations of oceanic tides. Pribl. Eq. Inv. Com,., No. 26, 1908, pp. 1-113. 4. Milne, J. The Peruvian earthquake of May 9th, 1877. Trans. Sets. Sac. Japan, vol. 2, 1880, pp. .50-96. 5. Platania, G. II maremotto dello Stretto di IVIessina del 28 dicembre 1908. Boll. Soc. Sis. Ital., vol. 13, 1908, pp. 369-458. 6. Wharton, W. J. L. On the seismic sea-waves caused by the eruption of Krakatoa, .August 26th and 27th, 1883. The Eruption of Krakatoa and Subsequent Phenomena (edited by G. J. Symons). 1888, pp. 89-150. CH. VI] SEISMIC SEA- WAVES 91 earthquakes, the effect on ships is as if they had struck on a rock or grated over a reef, and the surface of the water is strongly or tumultuously disturbed. The principal value of sea- quakes consists in the light which they throw on the distribution of submarine earthquakes (sect. 184). In the present chapter, waves of the second and third classes of seismic origin will alone be considered. They will be distinguished here as seismic sea-waves and seismic seiches. 93. Frequency of Seismic Sea-Waves. The number of earth- quakes accompanied or followed by seismic sea-waves is appa- rently small. In Mallet's catalogue of recorded earthquakes*, 47 earthquakes during nearly three and a half centuries (1501- 1842) are described as accompanied by sea- waves. In about twelve centuries (684-1897), 35 sea-waves were recorded in Japan in connexion with earthquakes. In all these cases, how- ever, the sea-waves were of considerable magnitude. The total number must be far larger. Though it is at present impossible to give exact figures, it seems probable that the proportion of submarine earthquakes accompanied by sea-waves does not differ greatly from the proportion of inland earthquakes that are accompanied by the formation of fault-scarps. 94. Nature of Seismic Sea- Waves at Places near the Epicentre. Soon after the earthquake, at intervals varying from a few minutes to half an hour or more, a great sea-wave approaches the land in one long imbroken ridge. As it passes into shallow water, the front of the wave becomes steep, then impending, and finally it breaks and sweeps over the loAv-lying ground. The general testimony of observers is that the first movement on the shore is a retreat of the water, which lasts for 5 or 10 minutes or even longer before the arrival of the waves. In some cases, as in the Riviera earthquake of 1887, this observation is con- firmed by the records of neighbouring tide-gauges. With regard to the height of the waves, personal testimony is usually mitrustworthy. At the same place, estimates may vary from 20 to 80 feet, and it is uncertain whether they refer to the height of the crest above the previous level of the sea or above the trough of the succeeding wave, that is, to the amplitude or the range of the waves. There can be no doubt, however, that * licj). lirit. Ass., 1852, \)\i. 1-170; 185:5, pp. 118-212; 1854, pp. 1-320. 92 SEISMIC SEA-WAVES [ch. the height is often considerable, and estimates can hardly be regarded as excessive which attribute a height of 30 feet to the sea- waves of the Alaskan earthquake of 1899, or a height of 50 or 60 feet to those of the Iquique earthquake of 1877 or the Lisbon earthquake of 1755, or 93 feet to the Sanriku (Japan) earthquake of 1896. In the Messina earthquake of 1908, which was far less intense than any of these earthquakes, Platania measured the heights of the waves from remains left on the walls or on the ground, and found that the maximum height on the Sicilian shore was 28 feet, and, on the opposite Calabrian coast, 35 feet between Pellaro and Lazzaro*. The height of the waves, however, does not depend entirely on the violence of the shock. It is governed jjartly by the in- clination of the sea-bed in the neighbourhood of the shore, partly by the form of the coast, being naturally great at the head of a graduall}^ narrowing bay. In the Messina earthquake of 1908, the greatest heights were attained at some distance from the epicentral districts. In the Valparaiso earthquake of 1906, no sea-Avaves were noticed at Valparaiso itself owing to the great depth of water along the coast, while, at Kushimoto in Japan (10,937 miles from Valparaiso) the total range of the movement was 4 inches. The first sea-wave is usually followed by several others, as a rule, but not always, of less magnitude. After the Concepcion earthquake of 1835, three waves swept over the town; in the Japanese earthquake of 1854, there were seven or eight waves; in the Iquique earthquake of 1877, the sea broke eight times over both Iquique and Arica; the Lisbon earthquake of 1755 was followed by 18 waves at Cadiz. One of the most carefully observed series of sea-waves were those associated with the Sanriku (Jajjan) earthquake of Jime 15, 1896. The shock was felt at Miyako at 7.32 p.m. About 7.50, the sea began to retire; it then rose imtil about 8,0 and again retired. At 8.7, the largest wave invaded Miyako, and other large waves swept the shore at 8.15, 8.32, 8.48^ 8.59, 9.16 and 9.50 p.m. The intervals between successive waves were thus 7, 8, 17, 16, 11, 17 and 34 minutes; from Avhich it may be con- cluded that the waves had periods of about 16 and 8 minutes. * Platania, pp. 371-428. VI] SEISMIC SEA-WAVES 93 The period of the oscillations commonly observed in the Bay of Miyako in ordinary weather is somewhat different, namely, 21 minutes*. 95. Effects of Seismic Sea-Waves. A few examples of the power of seismic sea-waves may be given. The waves referred to in the last paragraph destroyed the town of Kamaishi and swept many of the houses into the sea. Two schooners were left among the ruins, one of them 200 yards from the shore. On the east side of the Bay of Yakutat in Alaska, mature trees were wrecked by the sea- waves of 1899, the beach presenting a wild, almost impenetrable, tangle of uprooted, broken, twisted and shattered trunks. The Messina sea-waves of 1908 wrought most damage on the Calabrian coast of the Straits. At Reggio, a block of concrete, 8i feet long, 8 feet wide and 4' feet thick, was torn from the pier and carried more than 20 yards by the waves. The sandy shore between Pellaro and Lazzaro was also swept away, in places to a width of more than a hundred yards t. 96. Distances traversed by Seismic Sea-Waves. The earth- quakes on both sides of the Pacific Ocean generate sea-waves that are observed on the opposite coasts. The waves of the Japanese earthquake of 1854 were recorded at San Francisco (5089 miles) and San Diego (5593 miles); those of the Sanriku earthquake of 1896 at the former i)lace (4787 miles). Much greater are the distances traversed by the waves due to South American earthquakes. The Peruvian earthquakes of 1868 and 1877 were registered by tide-gauges at Hakodate in Japan (10,315 miles), those of the Ecuador earthquake of 1906 at P'ukahori in Japan (9992 miles), and those of the Valparaiso earthquake of 1906 were recorded at Kushimoto in Japan (10,937 miles). The Krakatoa sea-waves of 1883, though not of seismic origin, were recorded at Havre (12,867 miles). The great distances traversed by sea-waves are of course due to the fact that the waves diverge })ractically in two dimensions only. 97. Nature of Seismic Sea- Waves at great Distances. As the sea-waves diverge from the epicentre, they diminish rapidly in amplitude, becoming long low swells, 100 or 200 miles in length. * IIoiKhi, pp. 90-91. t K. S.Tarr and L. .Martin, V. S. Geol. Surv., Prof. Paper No. <)9, 1912, pp. M\-^H; K. Oinori. Bull. Eif. Iiiv. Com., vol'. :J, 1909, |)p. 41-42. 94 SEISMIC SEA-WAVES [CH. VI At distances of a few thousand miles, they no longer retain the appearance of sea-waves, and approximate to that of tidal waves of short period, from 20 to 30 minutes in duration, the sea slowly rising and sweeping up the beach and then as slowly falling. It is no doubt from this gradual rise and fall that seismic sea-waves have been erroneously called tidal waves. For in- stance, in Hakodate Bay (Japan), the Iquique sea-waves of '6 p.m. 8 10 Mid. Fig. 42. Sanriku sea- waves (1896) recorded at Ayukawa. 1877, after travelling 10,315 miles, appeared as miniature tides with a period of about 20 minutes, a maximum amplitude of nearly 8 feet, and a total duration of several hours*. 98. Nature of Seismic Seiches. The nature of seismic seiches, as recorded by tide-gauges, Avill be evident from Figs. 42, 43. Fig. 42 is a reproduction of part of the record of the Sanriku waves of 1896 at Ayukawa, 168 miles from the epicentre. Fig, 43 represents the seiches due to the Iquique earthquake of 1877, as recorded at San Francisco, 5240 miles from the epicentre. The records differ of course chiefly in the amplitude of the * Honda, pp. 84-86. 96 SEISMIC SEA-WAVES [ch. oscillations and in their respective ratios to the amplitude of the tidal wave. The San Francisco record (Fig. 43) shows the principal features of seismic seiches: (i) their long duration, in this case for more than tw^o days; (ii) the prevalence of certain periods in the oscillations, the average periods of comparatively regular series being 17-3, 27-8, 34-3 and 47-4 minutes, the most conspicuous being 34-3 minutes and the octave 17-3 minutes; and (iii) the relation of one of these periods (34-3 minutes) with that of one of the periods of oscillation (38-48 minutes) of the water of San Francisco Bay. 99. Periods of Seismic Seiches. The similarity of seismic sea-waves in bays to the seiches observed in lakes was pointed out by Omori in 1901, when he discovered that the waves re- corded in each bay, whether excited by earthquakes or by storms, have their own proper period or periods. He inferred that each enclosed portion of the sea is virtuall}^ a fluid pen- dulum, the oscillations of which are fixed in j^eriod by the form and dimensions of the bay. Elaborate observations on marine seiches have since been made by K. Honda and his colleagues. Some examples of the regular periodicity of seismic seiches may be given. Hakodate Bay (Japan) is semicircular in form. The periods of the most conspicuous seiches vary from 45-5 to 57-5 minutes and from 21-9 to 24-5 minutes, the former corresponding to the fundamental oscillation of the bay, the latter to its lateral oscillation. The periods of the Sanriku seiches (1896) were 18-8. 39-5 and 57-5 minutes, and those of the smaller seiches dvie to an earthquake of 1897 were 22-1 and 45-5 minutes. The periods of the Ecuador seiches of 1906 were 21-9 and 40-9-49-2 minutes, and those of the Valparaiso seiches of the same year 22-1 and 48-0-53-0 minutes. At Honolulu, the average periods of the Krakatoa seiches (1883) were 27-7 minutes, of the Sanriku seiches (1896) 23-4- 26-0 minutes, of the Ecuador seiches (1906) 24-8-26-8 minutes, and of the Valparaiso seiches (1906) 26-2 minutes. All of these periods agree closely with that of the fundamental oscillation of the inlet leading to Honolidu. The Messina seiches of 1908 were recorded by nine tide-gauges in the Mediterranean, and, with one exception, their periods VI] SEISMIC SEA-WAVES 97 corresponded closely with those of seiches due to meteorological causes *. The cause of the stable periodicity of marine seiches in any bay has been explained by Omori and by Honda and his colleagues. Considering first, for simplicity, the case of a rect- angular bay of length / and uniform depth h, if waves approxi- mately of period UjVgh be propagated from the ocean directly up the bay, such waves will be reflected at the head of the bay, and, by the interference of the incident and reflected waves, a stationary wave will be formed with its joop at the head, and its node at the mouth, of the bayf. Xow, if waves of different periods proceed from the ocean towards the shore, the wave with a period coinciding nearly with that of the oscillation of the bay water Avill excite the most energetic oscillations of that water. Thus, the bays on a coast-line may be compared with a series of resonators, each of which selects, and resounds to, the note of its proper period from a chaos of sounds outside. Besides the fundamental or uninodal oscillation referred to, oscillations with two, three, or more, nodes are possible, the periods of such oscillations being respectively one-third, one- fifth, etc., of that of the fundamental oscillation. In some cases, also, the lateral oscillation of the bay is possible, as in San Francisco Bay, in which the frequently observed period of 34*3 to 41-2 minutes is due to lateral oscillations between the West Berkeley and Sausalito sides of the bay. In bays of regular shape, the position of the nodal line is determinate. In bays of complicated form, however, several nodal lines are possible, to each of which there will correspond a special period of the seiches. In such a bay, the seiches may thus have several different periods; and, as in the case of the Iqviique seiches at San Francisco (Fig. 43), the periods may at times undergo changes from one epoch to another J. * 1). Kikuclii, I'uhl. lui. Inv. Cinii., No. li), 1904, p. '20; Honda, pp. l!)-20, (M)-02; IMatania, pp. J.2:j-428, 442-446. f The distance between successive nodes of the wave is thus 2/, an: Diitton, pp. 289, 291. t C. Davison, Geol. Mag., 1893, pp. :i56-:i(iO; J. Milne, Seis. Journ., vol. 1, 1893, p. 134; Dutton, pp. 283-295; Oldham, pp. 97-99. 106 SECONDARY EFFECTS OF EARTHQUAKES [ch. and the inconstancy of their direction. They may be divided into five classes: (i) rift-fissures, due to the differential move- ment along a fault or rift; (ii) bluff -fissures, formed on hill-sides or in horizontal ground near trenches; (iii) plain fissures formed in horizontal ground but unconnected with trenches; (iv) hill- foot fissures, formed in alluvial plains at the foot of hills; and (v) fault-block fissures, consisting as a rule of two parallel fissures with depressed ground between. Rift-fissures have been described in Chapter V (sect. 76), and will not be referred to further in this chapter, as they are not secondary effects of the shock*. 112. Bluff-fissures. Fissures on hill-sides are merely small incipient landslips, and both owe their origin to the same cause. The fissures run parallel or nearly so to the strike of the sub- jacent rock-surface, and there is usually a slip of the ground on the downward side of the fissure. In the case of incipient earth-slumps, the fissure is horseshoe-shaped. One formed during the Andalusian earthquake of 1884 was about 2 miles long, from 10 to 50 feet wide, and of great depth. It was bordered by in- numerable minor fissures, some parallel and others perpendicular to the main fissure. Houses in the enclosed tract were shifted as much as 30 yards within a month after the earthquake. 113. On more or less level ground, bluff -fissures occur near the banks of rivers, or the sloping sides of reservoirs or tanks, and on elevated roadways and river-embankments. They are seldom straight for any distance, following the curve of the river-banks, to which they are nearly always parallel. As a rule, they are arranged in concentric curves, being closest to- gether near the river-banks, where the distance between suc- cessive fissures may be only a foot or two, but more frequently ranges from 10 to 15 feet. In length, they ^-ary greatly. Some are only a few yards long, the majority attain a length of a few hundred feet, some even a few hundred yards, while a length of a mile is exceptional. They are often, however, arranged in series, overlajDping one another en echelon, so that the total length of such a series may be considerable. Their width is also variable. In the Charleston earthquake of 1886, they were seldom more than an inch wide except near river-banks. In * Button, pp. 280-281. VII] SECONDARY EFFECTS OF EARTHQUAKES 107 other earthquakes, they may be 2 or 3 feet, or as much as 5 feet, wide. The depth is hmited by that of the superficial layer of alluvium, and rarely, it would seem, exceeds 15 or 20 feet. The origin of bluff-fissures is evident from their arrangement and distribution. They are clearly due to the settling of the ground near the unsupported margin of a trench or channel, and thus are merely incipient or undeveloped forms of land- slips*. 114. Fissures unconnected with Excavations. While the exist- ence of a neighbouring trench is essential to the occurrence of bluff-fissures, other fissures are occasionalh^ formed in flat alluvial ground far removed from any excavation, but only with earthquakes of great intensity. They differ from bluff- fissures in their greater magnitude and in the approximate con- stancy of their direction throughout a large area. During the New Madrid earthquake of 1 SI 2, many such fissures were formed in the alluvial plain of the Mississippi. In all, there was a ten- dency towards arrangement in parallel lines, the average direc- tion being about N. 30° E. When the earth-lurches in this district took the form of long low undulations, the fissures were usually aligned in the same direction. In the Assam earthquake of 1897, the fissures were found in many cases to run parallel to raised roads or embankments and on cither side of them. There can be little doubt that the fissures here considered owe their origin to the visible waxes which traversed the central district, (i) The formation of the fissures has been seen to coincide with the passage of the waves. During the New Madrid earthquake of 1812, the earth is said to have rolled in waves 3 feet high with visible depressions between, the waves finally breaking and leaving a scries of parallel fissures. During the Assam earthquake of 1897, fissures were also seen to open as one wave had passed and to close up again as another arrived, (ii) As already noticed, some of the fissures of the New Madrid earthquake were parallel to the undulations of earth-lurches. Thus, though many fissin-cs so formed may have closed up again, * R. Mallet. The Great Neaj/olilan Earthiinithc of 18.>7, vol. 2, 18()2, j)p. ;j(i'2-:U)5; J. Milne, Seis. Journ., vol. 1, 189;J, p. VAi; Dutton, pp. 280- 281; C. S. Middlemiss, Mem. Geol. Surv. India, vol. 38, 1910, p. 122; Oldham, pp. 10, 13, 21 ; Lawson, pp. 401-402; Fuller, pp. 47-56. 108 SECONDARY EFFECTS OF EARTHQUAKES [ch. others must have continued open, especially when the com- pression of the ground remained permanent in the form of earth-lurches *. 115. Hill-foot Fissures. During the Assam earthquake of 1897, fissures were formed at the foot of the Khasi and Garo hills wherever the alluvium of the plains extended up to the hill-side. It was found that the alluvium had separated from the hills and had dropped almost vertically through a distance of 1 to 5 feet, as at he (Fig. 44), giving the appearance of a fault- scarp, except that it followed the windings of the hill-foot. Beyond this, a strip of the alluvium from 10 to 20 feet wide was slightly depressed, and, still farther, a band, d, slightly raised above the original level of the plain. Shortly after the Fig. 44. Hill-foot fissure in the Assam earthquake of 1897. earthquake, the varying height of the alluvium was rendered evident by the flooding of the plain for purposes of irrigation, the water covering the undisturbed surface but not the elevated strip. Oldham attributes the formation of hill-foot fissures to the repeated thrust of the hill and plain, the one against the other, during the shock, the alluvium lurching into a low ridge during the compression, while the drop and depression were formed during the return movements f. 116. Fault-Block Fissures. Fault-block or compound fissures have been observed in the central districts of the New Madrid earthquake of 1812 and the Assam earthquake of 1897, espe- cially in those areas in which a layer of clayey alluviimi overlies one of waterlogged sand. They consist of a pair of parallel fissures between which the ground has sunk (Fig. 45). Unlike * Fuller, pp. 47-57; Oldham, pp. 20, 26, 89-90; R. Mallet, Rep. Brit. Ass., 1850, p. 52. t Oldham, pp. 92-93. VII] SECONDARY EFFECTS OF EARTHQUAKES 109 bluff-fissures, fault-block fissures are often straight for con- siderable distances. In some cases, they are so regularly formed that they might easily be mistaken for artificial trenches, the steep sides and flat bottoms resembling those of canal excava- tions. In the New Madrid earthquake, the parallelism of the fault-block fissures is generally very marked, groups of two to five or more long, straight and parallel canal-like depressions being not unconmion. The spaces between them are usually much greater than between ordinary fissures, amounting ta several hundred feet and occasionally to half a mile. In length, they are also greater, the average being from 300 to 500 feet, though lengths of half a mile or more have been observed. In the Assam earthquake, the depth of the depressed region be- tween the fissures was about a foot or 18 inches. In both earth- Fig. 45. Fault-block fissures in the New Madrid earthquake of 1812. quakes, little or no sand was extruded through the fissures, and thus the undermining which allowed of the subsidence of the fault-block must have taken j^lace by the extrusion of the quicksand through more distant fissures or by its creep into the neighbouring rivers*. Effects ox Underground-Water 117. Effects on Springs and Wells. The general effect of an earthquake is to raise tiie level of water in wells and to increase the flow from springs. The effect is, however, by no means uniform; for, with the same earthquake, there may be a diminished supply from some springs while others cease to flow. In some cases, the change of level in wells is of brief duration, passing away as soon as the earthquake is over; but, as a rule, it lasts from a few hours to a week or more. Before the Colchester earthquake of 1884, the water-level at the Colchester water- works had been sinking. The shock, which was one of the strongest known in this country, caused a rise of 7 or 7^ feet. * Oldham, pp. 92-93; Fuller, pp. 47-58. 110 SECONDARY EFFECTS OF EARTHQUAKES [ch. The level gradualh^ declined, but remained above the normal height for about six months. In connexion with this subject, some interesting observations have been made by F. H. King on the level of the ground-water in a well in Wisconsin, U.S.A. The well is 140 feet from the nearest railway-line. The diagrams furnished by the gauge showed frequent sharp short-period curves denoting a rise in the level of the water, and these curves were found to be associated with the movement of trains past the Avell. The most marked rises were produced by heavily loaded trains which moved rather slowly. The diagram indicated a rapid but gradual rise of the water, followed by a slightly less rapid fall to the normal level, the movement beginning shortly after the passage of the train and amounting to about one-tenth of an inch. Somewhat similar observations have occasionally been made on the level of well-water during earthquakes. There are thus two distinct movements in the underground water associated with earthquakes, (i) a somewhat sudden rise in the level of the water at the time of an earthquake, followed by a rapid return to the normal level, and (ii) a rise in the water, returning to the original level after the lajase of weeks or months. The causes of the movements are not entirely clear, but it is probable that the short-period variations are due to the tran- sitory extrusion of the caj^illary water from the interstices of the soil; while the long-joeriod variations are caused by a tem- porary widening or closing of the fissiu-es through which the water circulates*. 118. Extrusion of Water from Fissures. The extrusion of tvater from fissures is a common phenomenon of great, and even moderately strong, earthquakes. It does not, however, take place from all fissures, but is usually confined to certain belts, cliiefly no doubt to those in which the surface alluvium overlies a bed of water-bearing sand. The water is often ejected with considerable force, carrying with it sand and other material from some depth. In several earthquakes, it has been seen to rise in continuous columns or * R. INIeldola and W. White, East Anglian Earthquake of April ^^.nd, 1884, pp. 155-162; F. H. King, U. S. Department of Agriculture, Weather Bureau, Bull. No. 5, 1892, pp. 67-69; C. E. de Ranee, Nature, vol. 30, 1884, p. 31. VII] SECONDARY EFFECTS OF EARTHQUAKES 111 jets to heights of from 2 to 4 feet, while stray splashes, as indicated by the sand left on trees, have reached a height of 13 feet or more. The ejection continues for several minutes, and in some cases for hours, after the earthquake. The total amount of water extruded may therefore be very great. After the Charleston earthquake of 1886, every stream-bed, though usually dry in summer, carried off the ejected water. During the New ^ladrid earthquake of 1812, the water forced its way through the surface-dei)osits, '-blowing up the earth with loud explosions." The quantity extruded was enormous, one tract many square miles in area being covered to a depth of 3 or 4 feet. The extrusion of water and sand is due to the compression of the layer of water-bearing sand between the alluvivnn and the imdcrlying beds. This compression may be effected partly by the shock itself, and, in some cases, partly by the upward thrust of the underlying beds. The continuance of the extrusion after the shock is over is probably due to the differential settling of the fissured alluvium at the surface*. Sand-Vents 119. The extrusion of water, accompanied by sand, is of two general types, (i) the more common form of violent ejection in jets described in the preceding section, and (ii) the quiet ex- trusion of water. With each type is associated its peculiar form of sand-vent; with the former, the sand-craters common in all great earthquakes, and with the latter the rare form of sand- blows, sand-sloughs, etc., manifested during the New Madrid earthquake of 1812. 120. Sand-Craters. A typical sand-crater is shown in Fig. 46, which represents one of those formed during the Assam earth- quake of 1897. It consists of a saucer-shaped crater of sand, with its rim slightly raised above the original level of the ground. Sand-craters are of all sizes from about a foot to 20 feet or more in diameter, and several feet in depth. They occur only in the central area of the earthquake, and in those parts of it in which there is a water-bearing layer of sand at a short distance * H. Mallet and T. Oldham, Quart. Juurn. Geol. Soc, vol. 28, 1872, pp. 2.59-2(50, 2GG-209; Oldliam, pp. 13, 15, 20, 101-104; Lawson, p. 403; Dutton, pp. 281-282, 289; Fuller, p. 76; K. Meldola and W. White, East Anglian Karlh(iuakc of April 'I'lnd, 1884, p. 77. 112 SECONDARY EFFECTS OF EARTHQUAKES [ch. below the surface. In some places, they are close togetheiv almost overlapping; in others, they are several yards apart. After the Assam earthquake of 1897, 52 craters were counted in a strip of land near Maimansingh containing about one-eighth of an acre. The amount of sand ejected from the craters is often considerable. In the central district of the Charleston earth- quake, many acres were covered with sand, which was 2 feet or more in thickness near the orifices and thinned out toAvards the margins. In this earthquake, the sand, as a rule, came from a few feet only below the surface; in the Californian earthquake Fig. 46. Sand-vent in the Assam earthquake of 1897. of 1906, it was in one place the same as that pierced by wells at a depth of 80 feet. The formation of sand-craters is due to the extrusion of water and sand, as described in the last section. By the rapid outflow of the water, the fissures at one or more points are worn into round channels of considerable size, which are still further en- larged at the surface by the back-rush of the water into the fissure. As the force of the current begins to slacken, a circular mound of sand is deposited round the orifice, within the rim of which a hollow is formed by the water rushing downwards. Gradually, the orifice becomes choked with sand, through which the water filters back ; and thus the bottom of the crater, instead of narrowing almost to a point, is broad and nearly flat, though slightly hollowed*. * Button, pp. 281, 284; Oldham, p. 20; Lawson, p. 403; R. Mallet and T. Oldham, Quart. Journ. Geol. Soc, vol. 28, 1872, pp. 266-269. VII] SECONDARY EFFECTS OF EARTHQUAKES 113 121. Sand-Blows and Sand-Sloughs. The typical sand-blow of the New Mad id earthquake is a nearly circular patch of sand, from 15 to 18 feet in diameter, and 3 to 6 inches high, with a low rounded profile with concave slopes, but without any trace of a central depression. The large sand-blows are 100 feet or more in diameter and about a foot in height. Some are elongated in form, about 200 feet long and 25 to 50 feet wide. They are confined to the low alluvial lands of the Missis- sippi basin, but are absent from the immediate neighbourhood of the river. In places, they merged into one another, so that the whole surface was covered with a continuous sheet of sand. Sand-sloughs are low, someAvhat ill-defined, ridges of sand parallel to one another, and alternating with shallow troughs in which water is collected in long narrow pools. The ridges and depressions, as a rule, are only a few inches in height or depth. Sand-sloughs are always found on low ground, but very rarely near the Mississippi. Both sand-blows and sand-sloughs are formed by the quiet extrusion of water and sand, caused probably by the unequal settling of the alluvial deposits into the water-bearing stratum below. Their absence from the higher lands appears to be due to the greater thickness of the deposits above the quicksand in those regions; and their absence from the neighbourhood of the Mississippi to the water and quicksands flowing laterally into the river*. Rise of River-Channels, etc. 122. During the Assam earthquake of 1897, river-channels, tanks, wells, etc., were filled up over a large area, partly perhaps by the abundant outpouring of sand, but chiefly by the forcing- up of the bottom. Trenches or excavations of any kind are places of special weakness in the surface layer, and the sand underlying it would be forced up readily through such openings. Many river-channels from 15 to 20 feet deep were almost com- pletely obliterated in this manner, so that the rivers aftei wards (lowed on shallow sandy beds. Aiiother effect, due to the same changes, was a sudden rise in the rivers of from 2 to 10 feet, which took place immediately * Fuller, pp. 70-89. D.M.S. 8 114 SECONDARY EFFECTS OF EARTHQUAKES [ch. after the earthquake, but passed off in a day or two. At Gauhati, for instance, the Brahmaputra rose more than 7| feet within three-quarters of an hour of the earthquake, but retiu'ned to its normal level in 2| days. At Goalpara, the first rise in the same river was not less than 10 feet. This rise may have been partly due to the great quantity of water extruded from the sand-vents, but it was too sudden to be entirely explained in this manner. It was probably caused by the forcing-up of the river-bed, more in some parts than in others, so that the water was partially ponded back by barriers. Being composed of sand, such barriers would be rapidly scoured away and the river would return to its normal level*. Effects of Earthquakes on Glaciers 123. The effects of earthquakes on glaciers have been studied in the case of one series of earthquakes, those of Alaska in 1899. To the north of the epicentral district of Yakutat Bay lie the lofty St Elias and other ranges, from which many glaciers descend towards the coast. The effects of the earthquakes on these glaciers are of two kinds. (i) An immediate effect of the earthquakes was the shatterirg of the glacier-ice over a wide area, a shedding of icebergs from the glaciers which descend into the sea, and a consequent re- cession of the glaciers. Even in Glacier Bay (150 miles from Yakutat Bay), the front of the well-known Muir Glacier was so shattered that the inlet into which the glacier flowed was in- accessible to steamships for eight years. During the 13 years from 1894 to 1907, the total retreat of the Muir Glacier was 8| miles, and of the Grand Pacific Glacier 8 miles, the retreat in both cases l)eing largely, but perhaps not entirely, due to the earthquakes of 1899. (ii) A second and far more important effect of the earth- quakes was due to the vast snow-avalanches that fell from the Alaskan mountains and accumulated in the glacier reservoirs. In 1899, all the Yakutat Bay glaciers were retreating, and this retreat continued for several years longer, being no doubt aggra- vated by the shedding of icebergs at the time of the earthquakes. Then, one by one^ the glaciers began to advance, the shortest * Oldham, pp. 104^108. VII] SECONDARY EFFECTS OF EARTHQUAKES 115 first at some time before 1905, and then in turn those of in- creasing length from 1905 onwards, the longest of all being still unaffected in 1910. The changes undergone were similar in all the glaciers. The advance was abrupt and spasmodic; the sur- face, previously smooth and easily traversed, was transformed into a wilderness of pinnacles and crevasses ; the glaciers thick- ened at their lower ends; and finally, after advancing several hundred yards in ten months or less, the effects of the increased supply of snow were spent and the glaciers returned rapidly to a stagnant condition*. * R. S. Tarr and L. Martin, U. S. Geol. Siirv., Prof. Paper No. 69, 1912, pp. 51-61. 8—2 CHAPTER VIII POSITION OF THE SEISMIC FOCUS 124. The main subject of all earthquake-investigations is the position of the seismic focus. If the focus were a point, the elements required would be the latitude and longitude of the epicentre and the depth of the focus below the surface. In many catalogues, the latitude and longitude of the epicentre are given, the reference in such cases being to the position of a point which may be regarded as the centre of the epicentral tract or of a point which coincides approximately with that in which the shock attained its greatest strength. Even Mallet, to whom Ave are indebted for the term seismic focus, realised that the dimensions of the focus must be considerable. His investigation of the Neapolitan earthquake of 1857 — the first attempt to grapple seriously with the problem before us — led to the con- clusion tiiat the focus was a fissure 10 miles long in a horizontal direction, and in depth varying from 4f to 8^ miles. The realisation that the focus possesses finite magnitude in- creases the complexity of the problem. We have not only to determine the mean position of the epicentre ; we have also to ascertain the form and dimensions of the epicentral tract. To a great extent, both are problems within our present grasp, and the method of their solution will now be described. With regard to the depth of the focus, our methods are less accurate and, consequently, our knowledge is uncertain. Some of the methods which have been suggested will be explained below. The present chapter will be confined to the case of earthquakes in which the epicentral tract can be studied or is near at hand. The determination of the origin of distant and unfelt earthquakes will be considered in the next chapter (sects. 162-166). Position of the Epicentre 125. In a great earthquake, in which the central tract is crushed and faulted, the determination of the epicentre is usually a simple matter. The dislocation of the ground and the scarp CH. VIII] POSITION OF THE SEISMIC FOCUS 117 of the faulted rock are more or less permanent traces of a portion of the epicentral tract. The whole of the area it may be more difficult to outline. The distribution of the after-shocks, as in the Mino-Owari earthquake of 1891, may indicate a fault-displace- ment of which no signs are left on the surface (sects. 208, 215). In a complex earthquake, like the Assam earthquake of 1897, Oldham used various lines of evidence — the course of the fault-scarps and rock-fissures, the folding of the crust, and the 118 POSITION OF THE SEISMIC FOCUS [ch. distribution of the after-shocks — and assigned to the epicentral area the form shown in Fig. 47, a vast region 200 miles in length and 6000 sq. miles in area*. In earthquakes unaccompanied by fault-displacements, we have to rely on other evidence, the methods which have been suggested depending on observations of the time, the direction of the shock, and its intensity, respectively. 126. Methods depending on the Time. The methods which depend on observations of the time of occurrence at five or more stations need not be described in any detail. They have been used in many earthquakes, but the results obtained are un- trustworthy, because (i) it has never been found possible to obtain sufficiently accurate records of the time; and (ii) even if the observations were accurate to the nearest second, different phases of a movement lasting for many seconds or several minutes Fig. 48. Omori's method of determining might be timed in different the epicentre. , ^ places. Omori's method, which depends on the duration of the pre- liminary tremor at three or more stations, is free from these objections. If y be the duration in seconds of the preliminary tremor at a place distant x kms. from the focus, Omori has shown (sect. 60), that, in earthquakes in which x is less than 1000, X and y satisfy the equation X = 7'21y + 38. If the durations of the prehminary tremor be, say, 40, 65 and 72 seconds at the stations A, B and C, respectively (Fig. 48), the distances of the focus from them must be 330, 510 and 550 kms. Assuming that the depth of the focus is small com- pared with these distances, it follows that the epicentre must lie on each of the three circles described with A, B and C as centres * Oldham, pp. 164^179. viii] POSITION OF THE SEISMIC FOCUS 119 and the above distances as radii, and therefore at or near the point in which the circles intersect. In practice, the circles do not pass through a point, but form a small triangle, the centre of Avhich ma}' be regarded as giving the position of the epicentre. The method is used in Japan for determining the epicentres of the submarine earthquakes which originate off the east coast of that comitry*. 127. Method depending on the Direction. The second method — that depending on observations of the direction of the shock- is the well-knoAvn method used by Robert Mallet, though suggested by John Michell in 1760. Mallet assumed that, during the passage of the wave, each particle of rock moves in a closed curve the longer axis of which is directed towards the focus, and, in practice, he regarded the curve as a straight line. He inferred that monuments, gate-posts and loose objects would fall, or be projected, in the direction of the wave-path at each place, and that fissures in the walls of buildings — and he laid stress on the necessity of the buildings being simple and sym- metrical in shape — would be formed along perpendicular lines. As such wave-paths would intersect in the epicentre, two lines of direction would be sufficient if the epicentre were a jDoint, while, if otherwise, a larger number would intersect two and two at different points within the epicentre and might thus plot out roughly its form and magnitude. Though simple in form, the method is by no means eas}' to apply. The first aspect of a ruined city, as Mallet found in studying the Neapolitan earthquake of 1857, is one of be- wildering complexity. Houses arc overthrown in every con- ceivable direction. There seems at first sight to be no governing law in their fall. It is only, as he remarks, by first gaining some commanding point, and by viewing the place as a whole, that any prevailing direction seems to stand out clearly from the surrounding confusion. If Mallet had measured the bearing of every suitable fissure and of the fall of every object, and had taken the mean of all the observations, he would probably have obtained a very close a])j)roximation to the jirincipal direction of the shock, as Omori did with the fallen stone-lamj)s at Tokyo (sect. 58). The method he adopted was apparently * K. Omori, Pubf. Ki/. Iiiv. Com.. No. 21, 1905, pp. -M-'M. 120 POSITION OF THE SEISMIC FOCUS [ch. to form a first mental impression of the mean direction and then to measure particular lines which seemed to agree most closely with that impression. The results of Mallet's investigation of the Neapolitan earth- quake are shown in Fig. 49. The line drawn from each place represents the direction of the shock at that place. As will be seen, these lines are not absolutely concurrent, though they show a marked tendency to converge towards the village of Fig. 49. Mallet's method of determining the epicentre of the Neapolitan earthquake of 1857. Caggiano. They intersect, indeed, at various points within an area indicated by the curve on the map — a line which Mallet regarded as the surface-trace of a fissure about 10 miles long, within which the earthquake probably originated. The method of directions, which Mallet used with such skill in this disastrous earthquake, has lately fallen into undeserved neglect. The extreme variability of the direction in any place cannot be overlooked, and solitary observations are not only valueless, but misleading. But, if the method be modified by the VIIl] POSITION OF THE SEISMIC FOCUS 121 substitution of the mean of a large number of measurements for a few scattered observations, it may still be foimd a useful instrument in the investigation of a destructive earthquake (sects. 58, 59)*. 128. Method depending on the Intensity. The first method depending on observations of the time, even if it could be applied with accuracy, requires a large disturbed area for its use. For ^^FW Fip. 50. Map of the Inverness earthquake of 1901 . the method of directions, a strong shock is also essential. The chief advantage of the third method is that it can be applied in earthquakes of every degree of strength. In violent shocks, the area of ruined towns and villages usually encloses, if it does not mark out, the epieentral area. In the weakest shocks, the disturbed area is itself so small that any place at which it is felt may be regarded without much error as coincident with the epicentre. * H. Mallet, The Great Seajjolilun Earthquake of 1857, vol. 2, 1862, pp. 235-247. 122 POSITION OF THE SEISMIC FOCUS CH. In these cases, the determination of the epicentre is only approximate. Greater accuracy is attained by the construction of a series of isoseismal hues, the centre of the innermost curve being either close to, or within, the epicentral area. Thus, as shown in Fig. 50, the epicentre of the Inverness earthquake of 1901 lies about 3 miles south-west of the city of Inverness; the principal epicentre of the Derby earthquake of 1904 is about 1^ miles east of Ashbourne (Fig. 51). Fig. 31. Map of the Derby earthquake of 1904. In the map of Fig. 51, the centre of the innermost curve is excentric with respect to the next surrounding isoseismal. In this earthquake, the shock consisted of two distinct parts. A shock of this type is known as a twin earthquake, and it will be seen later (sect. 243) that such earthquakes owe their origin to nearly simultaneous impulses in two detached foci. The more important focus is that which corresponds with the centre of the innermost isoseismal. In a few earthquakes, as in the Doncaster earthquake of 1905 and the Kangra earthquake of VIII POSITION OF THE SEISMIC FOCUS 123 1905 (Fig. 52), it is possible to trace the isoseismal lines sur- rounding both epicentres and therefore to assign their positions. 129. Form and Position of the Seismic Focus. A carefully drawn series of isoseismal lines reveals, however, much more than the position of the epicentre. As will be seen from the examples here given, the curves are usually elongated in form, and the distances between successive isoseismals are unequal on the two sides of the axes. These are necessarv results of a lono- lie of Mile 20 30 Fig. .52. Mai) '*f the Kangra earthquake of Apr. 4, 1905. seismic focus inclined to the horizon, such as a portion of a fault-surface within which one rock-mass slips over the other. From the forms and relative position of the isoseismal lines, we may deduce the elements of the fault, the displacement along which results in a given shock, (i) The direction of the fault is parallel, or nearly parallel, to the longer axes of the isoseismal lines, (ii) The fa\ilt hades towards the side on which the inner isoseismals are the farther apart, (iii) The fault-line passes close to the centre of the innermost isoseismal, and lies on that side of it from which the fault hades, (iv) The difference 124 POSITION OF THE SEISMIC FOCUS [ch. between the lengths of the axes of the innermost isoseismal gives a rough measure of the depth of the seismic focus. The Inverness earthquake of 1901 (Fig. 50) may be taken as an example of these methods. The innermost isoseismal (of in- tensity 8, Rossi-Forel scale) is 12 miles long and 7 miles wide, the longer axis being directed N. 33° E. The next isoseismal (intensity 7) is 53| miles long, 35 miles wide, with its longer axis directed N. 82° E. The distance between the isoseismals is 9 miles on the north-west side and 14 miles on the south-east side. From these measurements, it may be inferred that the mean direction of the originating fault is about N. 33° E., that the fault hades to the south-east, that the fault -line passes a short distance on the north-west side of the centre of the isoseismal 8, and that the length of the principal part of the focus was at least 5 miles. Now, the epicentral area is crossed by the great northern boundary fault of the Highland district, the line of which passes at a distance of somewhat less than a mile (Fig. 97) on the north-west side of the centre of the isoseismal 8. The fault hades to the south-east and, in the neighbourhood of the epicentre, its mean direction is about N. 35° E. It thus satisfied all the conditions implied by the seismic evidence*. Depth of the Seismic Focus 130. While the position of the epicentre can be determined with some approach to accuracy, much less confidence can be felt in our estimates of the depth of the focus. That the upper margin of the focus may sometimes coincide with the surface is obvious — the formation of a fault-scarp is evidence of this. It is clear, also, that, in most cases, the mean depth of the focus is not considerable, that it must be measured in terms of a few miles rather than of many miles, (i) The limited area over which many earthquakes are felt shows that, in such cases, the seismic focus is shallow. As will be seen later (sect. 139), the more rapidly the intensity of a shock declines outward from the epicentre, the less must be the depth of the focus, (ii) Earth- quakes can only be produced by fault-slipiDing in that outer * C. Davison, Beitr. zur Geoph., vol. 9, 1908, pp. 220-224; Quart. Journ. Geol. Soc, vol. 58, 1902, pp. 377-397. VIII POSITION OF THE SEISMIC FOCUS 125 portion of the crust which is so rigid that the rock under strain will break rather than bend, (iii) In those earthquakes which are traced to movements along known faults, the epicentre lies on the downthrow side of the fault and at distances from the fault-line not exceeding 1 or 2 miles*. 131. Methods depending on the Time. If the seismic focus were a point and the shock of brief diu'ation, accurate time- observations at five or more stations would give the position Fig. 53. Von Secbach's time-curve. of the epicentre and the depth of the seismic focus. Such estimates of the depth have been made for several earthquakesf . * For instance, in the Bolton earthquake of Feb. 10, 1889 (Fig. 29), the epicentre was 2 miles from the Irwell \'alley fault: the epicentres of the Inverness earthquake of Sep. 18, 1901, and its numerous accessory shocks lay between ,'„ nule and 1^ miles from the Great Glen fault (Fig. 97); the ei)iccntrc of the Malvern earthquake of 1907 was about a mile from the boundary fault of the Malvern Hills; the epicentres of the numerous Ochil earthcjuakcs of 1900-1914 lii- witliin 2 miles of the great fault which nuis along the southern margin of the Ochil Hills. In the case of the Bolton earthquake of 1K89, the hade of the fault is known, namely, 28°; as the point of maxinnun intensity of the shock is about 4k miles from the fault- line, it follows that the (le|)th of the focus was about ."JJ miles. f For instance, Hogljcn estimates the ilepth of the focus in the New Zealand earthquake of 1888 to be about 24 miles, in that of 1890 to be between 20 and .'JO miles, and in that of 1891 to be probably a little less than 10 miles {Traits. \nv Zeal. Inst., 1890, pp. 471, 477; 1892, p. 365). 126 POSITION OF THE SEISMIC FOCUS [CH. But little or no reliance can be placed on the results, for the focus is always of some size, time-observations are seldom accurate to the nearest minute, and in any case may refer to different phases of the motion. Von Seebach has suggested a graphical method of dealing Avith observations of the time. In the accompanying diagram (Fig. 53), O represents the focus, A the epicentre, and AM a horizontal line along the surface. The distance of any place of observation is represented by a length AM measured along this line. From the other end M, a perpendicular line MP is drawn of a length proportional to the difference between the observed times at the place and the epicentre, that is, to the difference between the distances OM and OA. The points corresponding to the other end P of this and similar lines would, if all the measurements were exact, lie on the hyperbola APQ, from the form of which it is possible to calculate the depth of the seismic focus*. In practice, the points so plotted fall into groups on either side of a curve which is drawn so as to pass as closely to them as possible. Von Seebach's method has been used in the investigation of a few earthquakes, with the following resultsf : Earthquake Authority Mean depth in miles, about Rhenish, 1846 Sillein, 1858 Mid-German, 1872 Herzogenrath, 1873 1877 von Seebach von Lasaulx it 24 16 12 7 17 * Taking the vertical and horizontal lines through O as axes of x and «/, let X yhe the coordinates of P, and c the depth of the focus. Let MP = X {OM - OA), where \ is a constant. Then X - c = \ (Vc- + y- - c), or x^ - XV - 2 (1 - X) ca- + (1 - 2\) c^ = 0, which is the equation of an hyperbola. The equation of the asymptote in Fig. 53 is a; - Xj/ = c (1 - X), the line passing through the seismic focus when X = 1, in which case the equation of the curve becomes x- - y^ = c^. t K. von Seebach, Das mitteldeutsche Erdbeben vom 6. Marz 1872 (Leipzig, 1873), pp. 132-175; A. von Lasaulx, Das Erdbeben von Herzogenrath am 23. October 1873 (Bonn, 1874), and Das Erdbeben von Herzogenrath am 24. Juni 1877 (Bonn, 1878). viii] POSITION OF THE SEISMIC FOCUS 127 Von Seebach assumed that the velocity of the earth-waves is constant at all dejiths below the surface. If it should vary with the depth, the time-curve, as A. Schmidt has shown, would be no longer hyperbolic, but concave upwards at and near the epicentre and afterwards convex. The tangent to the curve at the point of inflexion intersects the vertical through the focus at a point above the focus, and thus the depth of this point would give an inferior limit to the depth of the focus. Galitzin, making use of good observations of the time at stations near the epicentre, thus found the depth of the focus of the South German earthquake of Nov. 16, 1911, to be 5-9 miles, with a maximum error of 2-1- miles*. 132. Another method depending on time-observations has been suggested by Omori. If y be the duration in seconds of the preliminary tremor at a place near the focus and distant cT kms. from it, Omori has shown (sect. 60) that x and y in ordi- nary earthquakes satisfy the equation cT = 7-42?/. Thus, if the joosition of the epicentre be known, and if the distance of the focus be calculated from this equation, the depth of the focus is easily ascertained. Omori has applied this method to the earthquakes which accompanied the eruptions of the Asama-yama in central Japan. In 1911, these earthquakes were registered at Ashino-taira, the horizontal distance of which from the crater is 5 kms. The mean duration of the preliminary tremors was -96 second, giving 7-2 kms. for the distance of the focus. It follows that the mean origin must be about 6 kms. below the summit of the mountain or about 4 kms. {2\ miles) below its base. In 1911 and 1912, the earthquakes were registered at Yuno-taira, the horizontal and vertical distances of which from the crater are 2-3 and -6 kms. The mean duration of the preliminary tremors being -64 second, it follows that the mean focus was 4-8 kms. from the observa- tory, or 4-8 kms. beloM* the summit of the mountain or about 3-0 kms. (nearly 2 miles) below its basef. * A. Schmidt, Jahr. des Vereiii.s fiir vaterl. Xalurk. in Wiirtlemb., 1888, pp. 248-270; 19(K), pp. 2(K)-2:J2 (abstract in Suture, vol. 52, 1895, pp. G31- 63.'i); B. Galitzin, Compt. Rend, des Stances de la Com. Si.s. Perm., vol. 5, 1918, p. :y.i5. t F. Omori, Bull. Et/. Inv. Com., vol. 0, l'912, i)p. 127-128, 238. 128 POSITION OF THE SEISMIC FOCUS [ch. 133. Method depending on the Direction. This method, which is due to Mallet, depends on the assumptions (i) that it is possible to determine the direction of the wave-path at one or more points on the surface, and (ii) that the direction of the wave- path at any point coincides with that of the focus. If the focus were a point and if the position of the epicentre were known, one measurement of the direction of the wave-path would be sufficient to determine the depth of the focus. If, on the other hand, the focus were of finite magnitude, the intersections of the wave-paths at different pairs of points might give some idea of the depths of the upper and lower margins of the focus. Mallet sought to determine the direction of the wave-path by measuring the inclination to the vertical of fissures in fractured walls. He chose in all cases large walls, composed of bricks or short-bedded stones, and containing very few windows and doors. For instance, in the cathedral at Potenza, he found, from a series of nearly parallel fissures, that the mean angle of emergence was 23° 7'. The distance of Potenza from the epi- centre at Caggiano being 17 miles, the depth of the focus below the surface of the ground would be 17 tan 23° 7', or 7^ miles, or about 6| miles below the level of the sea. Mallet also measured the angle of emergence at 25 other places, the farthest being- Salerno, distant 40 miles from the epicentre, at which the angle was lli°. The calculated depths corresponding to the different angles of emergence lie between 3 and a little over 9 miles with a mean depth of about 6| miles *. Now, if the waves proceeded from such dejDths as these, there can be no doubt that, in their passage to the surface, they would traverse rocks of varying density and elasticity, and that at every bounding surface they would be deflected from their previous course. If the true direction of the wave-path at Potenza were 5° more or less than that given above, the corresponding depth of the focus below sea-level would be 8| or 5 miles. If the observation at Salerno were subject to the same error, the depth might be 11 J or 4, instead of 6|, miles. * R. Mallet, The Great Neapolitan Earthquake of 1857, vol. 2, 1862, pp. 248-251. VIIl] POSITION OF THE SEISMIC FOCUS 129 In the following table* are given the approximate depths of the foci of seven earthquakes, obtained by Mallet's direction- method : Earthquake Authority Mean depth in miles, about Neapolitan, 1857 Isohian, 1881 1883 Andalusiai", 1884 Kashmir. 1885 Riviera, 1887 Verny (Turkestan), 1887 Mallet Johnston-Lavis Mercalli Taramelli and Mercalli Jones Mercalli Mouchketow * . io| 134. In his investigation of the Yokohama earthquake of 1880, Milne applied the direction-method to determine the depth of the focus, but estimated the angle of emergence from the vertical and horizontal components of the motion as re- gistered by seismographs. The results of different observations varied within wide limits, but Milne considered the most pro- bable depth to be from 1| to 5 miles. In two other Japanese earthquakes (Feb. 7 and Apr. 30, 1897) Omori and Hirata esti- mated from seismographic records at Miyako that the depths of the foci were 5*6 and 9-3 miles. The same objection of course applies to these observations as to Mallet's. Possibly the angles of emergence in the Miyako records are determined with greater accuracy than those from the directions of fissures, but they depend on the further assumption that there was at the time no tilting of the ground f. * R. Mallet, Great Neapolitan Earthquake of 1857, vol. 2, 1862, pp. 240- 251 ; H. J. .lohnston-Lavis, Monogra/ih of the Earlh(iuakes of Ischia, 1885, pp. 71-74; G. Mercalli, L' isola iriscltia ed it lerreniolo del 28 luglio 1883, pp. 133-134; T. Taramelli and G. Mercalli, Mem. li. .Iccad. del Lincei, vol. 3, 188(>. i)|). 69-71 ; K. J. .Tones, liec. Geol. Surv. India, vol.* 18, 1885, p|). 224-225; (i. Mercalli, .inn. delV Vff. Cenlr. di Meteor, e Geodin., vol. 8, 1888, p|). 246-248; J. V. Mouchketow, Man. da Com. GdoL, vol. 10, p. 149. In addition to the above arc the follovvin, p. 211 ; Dutton, pp. 317-:J20; 1). K<;initis, .Inn. de G^ogr. (Paris), 189.^; S. Areidiaeono, Ann. deir Vff. Centr. Met. e Geod., vol. 1(», 1805, j). 7; A. Cavasino, Boll. Hoc. Sis. Ital., vol. 18, 1914, pp. 433-435; C. S. Middlemiss, Mem. Geol. Surv. India, vol. 38, 1910, pp. 331-331.; E. Oddone, Boll. Soc. Sis. Ital., vol. 19, 1915, pp. 190-194; M. Stuart, liec. Geol. Surv. India, vol. 49, 1918, pp. 187-188. 9—2 132 POSITION OF THE SEISMIC FOCUS [CH. Earthquake Authority Mean depth in miles, about Charleston, 1886 (two foci) Constantinople, 1894 Syracuse, 1895 Marsica (Italy), 1904 Kangra, 1905' Marsica, 1915 Srimangal (Assam), 1918 Button Eginitis Arcidiacono Cavasino Middlemiss Oddone Stuart 12 and 8 21 ^ between 21 and 40 6 8 or 9 138. The methods used for determining the depth of the focus have been described at some length, chiefly on account of the interest of the problem. The most accurate results are probably those of Johnston-Lavis for the earthquakes of Ischia, and, next to these, may be placed Omori's for the earthquakes of the Asama-yama. Estimates greater than 10 or 12 miles are based on scanty evidence. Of the remainder, the utmost that can be said is that they indicate, not the depth of the focus, but simply its order of magnitude. Fig. 55. Biagram illustrating the relative depths of the foci of two earth- quakes determined by the more or less rapid variations of intensity. 139. The principle involved in Button's method (sect. 135) gives, however, a test of the relative depths of the foci of two earthquakes, which is often useful. The curves in Fig. 55 re- present the intensities of the shock at different distances from the epicentre, the continuous line corresponding to a focus 2 miles in de23th, the broken line to one at a depth of one-quai'ter of a mile. They are drawn on the assumptions that the intensity VIII] POSITION OF THE SEISMIC FOCUS 133 of the shock at any point varies inversely as the square of its distance from the focus, and that the impulses in the two foci are such that the intensity at the epicentre is the same in each case. Thus, the more rapid the decline outwards in the intensity of the shock, the less is the depth of the focus. For instance, many Etnean earthquakes (sect. 221) destroy every building within a small zone and yet disturb areas of only 100 or 200 square miles. In Great Britain, the strongest earthquakes are just capable of causing slight damage to buildings, yet the average area disturbed by them is more than 65,000 sq. miles. We may infer with confidence that the foci of the Etnean earthquakes are very much shallower than those of British earthquakes. CHAPTER IX PROPAGATION OF EARTHQUAKE-WAVES 140. When a disturbance takes place in an isotropic solid — that is, a body which has the same elastic properties in all directions about a point — two principal types of Avaves are pro • pagated outwards from the source. In one, the particles vibrate along lines normal to the wave-front ; in the other, in the plane perpendicular to this direction. During the passage of the former waves, the solid is alternately compressed and rarefied ; during the passage of the latter waves, it is distorted. The waves are thus known as longitudinal or condensationdl or com- pressional waves and transverse or distortional waves, respectively. If p be the density of the solid, k the resistance to com- ])ression, measured by the pressure required to produce unit contraction, and n the modulus of rigidity, measured by the stress required to produce unit shearing-strain, and if m = ^ + |w, the velocities of the waves are given respectively by the ex- pressions \/{m/p) and ■\/{nlp). Since m is greater than n, it follows that the velocity of the condensational wave always exceeds that of the distortional wave*. * No branch of seismology has attracted more attention recently than the subject of this chapter. Among the large number of memoirs devoted to it, the following may be mentioned as especially worthy of attention : 1. Geiger, L. and B. Gutenberg. t)ber Erdbebenwellen. Nach. der K. Gesell. der Wissen. zu Gottingen, Math.-phys. Klasse, 1912, pp. 623-675. 2. Knott, C. G. (1). Reflexion and refraction of elastic waves, with seismological applications. Phil. Mag., vol. 48, 1899, pp. 64^97, 567-569. 3. (2). The Physics of Earthquake Phenomena (Oxford Univ. Press), 1908, pp. 156-258. 4. (3). The propagation of earthquake waves through the earth, and connected problems. Proc. Roy. Soc. Edin., vol. 39, 1919, pp. 157-208. 5. Milne, J. Reports of the Seismological Committee of the British Associa- tion, 1896-1913. 6. Oldham, R. D. (1). On the propagation of earthquake motion to great distances. Phil. Trans., 1900 A, pp. 135-174. 7. ■ (2). The constitution of the interior of the earth, as revealed by earthquakes. Quart. Journ. Geol. Soc, vol. 62, 1906, pp. 456-473. 8 . Omori, F. Horizontal pendulum observations of earthquakes at Tokyo. CH.ix] PROPAGATION OF EARTHQUAKE-WAVES 135 Reflection and Refraction of Earthquake-Waves 141. In passing from one medium to another possessing different wa^■e-moduli or different densities, both condensa- tional and distortional waves are modified. Each may be re- solved into four new waves — a reflected condensational, a reflected distortional, a refracted condensational and a refracted distortional, wave — though, under certain conditions, one or more of the waves may be absent. The number and types of such subsidiary waves depend p?rtly on the angle of incidence of the earthquake-wave, and partly on the relative values of the elastic moduli and densities of the two media*. Ptibl. Eq. Iiw. Com., No. 5, 1901, pp. 1-82; No. 6, 1901, pp. 1-181; No. 13, 1903, pp. 1-142; No. 21, 1905, pp. 9-102. 9. Reid, H. F. Instrumental records of the [Californian] earthquake. The Californian Earthquake of April 18, 1906 (edited by A. C. Lawson), vol. 2, 1910, pp. 59-142. 10. Turner, H. H. Reports of the Seismological Committee of the British Association, from 1914. 11. Walker, G. W. Modern Seismology (Longmans), 1913, pp. 37-82. 12. Wiechert, E. and K. Zoeppritz. Uber Erdbebenwellen. Nach. der K. Gesell. der Wissen. zu Gottingen, Math.-phys. Klasse, 1907, pp. 1-135. Experiments on the determination of the elastic constants of various rocks are described by H. Nagaoka in Publ. Eq. Inv. Com., No. 4, 1900, pp. 47-67; and by S. Kusakabe in Jourii. Coll. Sci., Imp. Univ. Tokyo, vol. 19, 1903, art. 6, and vol. 20, 1905, arts. 9, 10; also in Publ. Eq. Inv. Com., No. 17, 1904, pp. 1-48; No. 22 B, 1906, pp. 27-49. * Let p, p' be the densities of the two media, /», n and m' , n' their elastic moduli; let 6 be the angle of incidence (and reflection) of a condensational wave travelling in the former medium, 6' its angle of refraction, and (p, '«' > /./ " •> ^' — cosec- 6 = cosec-

53 119 938 1722 784 30 388 (594 306 60 1 012 1103 491 90 796 1454 658 120 942 1729 787 146 PROPAGATION OF EARTHQUAKE- WAVES [cii. corrected by him as fresh and more accurate observations became available, and tables were compiled giving the times of transit for different arcual distances measured in degrees. Milne's J 05 Q. \ \ \ \ \ \ \ \ \ \ V \ CO tables were slightly modified by Wiechert and Geiger in 1907. The latest tables are those given by H. H. Turner, depending mainly on the observations collected for the Seismological Com- mittee of the British Association. For the first 120 degrees, IX] PROPAGATION OF EARTHQUAKE-WAVES 147 they are here reproduced, the curves P and S in Fig. 64 re- presenting graphically the variations in the times of the initial primary and secondary waves with the distance. The line L is the corresponding curve for the long waves, as determined by the observations collected by O. Klotz*. A study of the table and time-curves leads to some imjiortant conclusions. 151. (i) Velocities of Earth- Waves. Denoting the mean velo- cities of the initial primary and secondary waves by V^ and V^ , we find, from the values of P and S for an arc of 3°, that, for the upper strata of the earth's crust Vy = 7-1 kms. per sec. and Fg = 4-0 kms. per sec. For greater distances, however, the table shows a continual increase in the values of V^ and F, with the distance, whether the distance be measured along a great circle or along a chord. Measured along a great circle, the values of V■^ for arcual dis- tances of 30=, m\ 90° and 120° are 8-6, 10-9, 12-6 and 13-6 kms. per sec; those of V^ for the same distances are 4-8, 6-0, 6-9 and 7-7 kms. per sec. Measured along a chord, the corresponding values of V^ would be 8-5, 10-4, 11-4 and 11-7 kms. per sec, and those of Fg 4-7, 5-8, 6-2 and 6-4 kms. per sec. Thus, in both primary and secondary waves, there is a marked increase in the velocity with the distance as measured along either the arc or the chord, ^^'e conclude, therefore, that the primary and secondary waves must travel along curved paths and that the velocities of both waves increase with the depth below the surface, from which it follows that the paths on the whole are concave towards the surface. On the other hand, it is clear, from the straightness of the curve marked L in Fig. 64, that the time varies as the distance measured along the great circle from the epicentre to the station; in other words, that the velocity of the long waves along the surface is constant. From the mean of a large number of observations, O. Klotz finds this velocity to be 230 kms. jx-r min. or 3-8 kms. per sect * J. Milne, Rep. Brit. Ass., 1898, p. 2'i:i; O. Klotz, Bull. Seis. Soc. Amer., vol. 7, 1907, pp. G7-71. Turner's table is {^iven in the circulars issued by the Brit. Ass. Seis. Com.; it is reprinted in full in Pror. lioi/. Soc. Edhi., vol. 39, 1919, p. 198. t Bull. Seis. Soc. Amer., vol. 7, 1907, pp. (>7-71. 10—2 148 PROPAGATION OF EARTHQUAKE- WAVES [cii. 152. (ii) Distance of Epicentre and Time at the Origin. Though the time-curves of individual earthquakes may vary sUghtly from Turner's mean time-curves, the value of S — P (that is, the duration of the primary phase) is practically constant for any given arcual distance. If, then, a seismogram provides definite readings for P and S, the corresponding figure for 6" — P in the table gives in degrees the distance of the epicentre from the station. It is obvious that determinations of the distance made at three widely separated observatories would fix the position of the epicentre (see sect. 164). Further, knowing the distance of the epicentre, we can find from the table the corre- sponding value of P and thus ascertain the time at which the earthquake occurred at the origin*. For instance, let P (the epoch of the first primary wave), as determined from a given seismogram, be 12 h. 6 m. 52 s., and let S — P be 674 seconds. From the table, we find that the epicentral distance is 93° (that is, 10352 kms,), and that P is 812 sees, or 13 m. 32 s. Thus, the time at the origin was 12 h. 6 m. 52 s. less 13 m. 32 s., or 11 h. 53 m. 20 s. 153. Nature of the Primary and Secondary Waves. We have seen that the motion of a distant earthquake is divisible into three well-marked phases, of which the primary and secondary waves pursue paths which penetrate the body of the earth, while the principal waves travel across its sm'face; and this division naturally led to the suggestion (by Oldham in 1899) that the primary waves consist of condensational, and the secondary waves of distortional, vibrations. Later observations, and especially those on the direction of the vibrations, on the whole support this view, though it is clear that condensational and distortional vibrations exist in both phases, the former pre- dominating in the primary waves and the latter in the secondary waves. It is obvious that, in the outer crust of the earth, neither condensational nor distortional Avaves could for long maintain their simple character without being split up into waves of both types as they passed from one rock to another (sects. 142-144). The records of seismographs near the epicentre of an earthquake * This time is usually denoted by the letter O, and the distance of the station from the epicentre by A. IX] PROPAGATION OF EARTHQUAKE- WAVES 149 cannot therefore be expected to show any separation of con- densational from distortional waves; and, indeed, this never takes place within a distance of about 10°, or 700 miles, from the epicentre. Since, at stations beyond this limit, the cha- racteristic triple phases begin to appear on seismograms, it is clear that the waves arriving there must, for part of their journey, have traversed some homogeneous material situated beloAv a comparatively thin layer. On re-entering or entering this layer, the waves which have become simplified in type must again, by mmierous reflections and refractions, become complex, although maintaining on the whole their condensa- tional character in the first series and their distortional form in the second. This explanation accounts for the nature of the two phases and for the greater velocity of the primary Avaves, but not for the continuity of the vibrations throughout each of the two early phases. Partly, the subsequent vibrations of each series may be due to internal reflection of the initial waves, as shown in sects. 158, 159, but such vibrations would be discontinuous, and the observed continuity and irregularity of the later movements is probably caused by repeated reflections and refractions at the bounding surfaces of the rocks which constitute the outer crust. 154. One other point AAith regard to these phases remains to be considered. While the weakness of the primary wa\'es seems to be the only obstacle to their registration at great distances, the secondary waves apparently die out at stations more than 110° or 120° from the epicentre. Their place in the series is not unoccupied, but the vibrations which follow the primary waves at such distances arc different from the typical secondary wa^es. Instead of beginning with a strong movement, with a maximum scton attained, followed by a rapid decline in strength, the substituted movement begins gradually, and there is no well- marked maximiuii. but r?thcr a succession of impulses. More- over, as G. \\. Walker points out (p. 42). the times at which they appear agree with the times at which waves reflected at or near the surface would reach the station (sects. 158, 159). Hut. whether or no this be the correct explanation of the move- ments, it is clear that the secondary waves cease to exist as such at distances from the epicentre greater than 110° or 120°. 150 PROPAGATION OF EARTHQUAKE- WAVES [ch. 155. Nature of the Long Waves. While there is general agree- ment among seismologists as to the origin of the primary and secondary waves, the nature of those which constitute the principal portion of the movement is somewhat uncertain. They are usually identified with surface-waves, known as Rayleigh waves, the existence of which was proved by Lord Rayleigh and H. Lamb. The velocity of the distortional waves in the outer rocks is known to be 4-0 kms. per sec. That of the Rayleigh waves, according to theory, is -9194 x 4-0, or 3-7 kms. per sec, a result which is in close agreement with Klotz's estimate of .3-8 kms. per sec. for the mean velocity of the initial long waves. Even if this identification be correct, it is still difficult to account for the great duration of this phase of the movement, and especially for that of the end portion or coda, which may last as many hours as the shock does minutes at the epicentre*. 156. Form of Seismic Rays. The determination of the velo- cities of the primary and secondary waves shows, as already mentioned (sect. 151), that these waves cannot travel along the surface or along chords to the various stations at which they are recorded, but rather along curved paths which, on the whole, are concave towards the surface. The exact form of these paths or seismic rays has been considered by several seismolo- gists, and especially by C. G. Knott, whose latest work is here described. The only assumptions which Knott requires in this work are : (i) that the earth is a sphere, (ii) that the earthquake originates so near the surface that the depth of the focus, in comparison with the radius of the earth, may be neglected, and (iii) that the elastic properties of the earth's substance depend only on the distance from the centre, and are therefore constant over any sphere concentric with the earth. If these assumptions be correct, every seismic ray which emerges at a given station must lie in the plane through the focus, the station and the centre of the earth. In accordance with the well-known law which governs all elastic waves, every ray must be such that the vibrations * Rayleigh, Proc. Math. Soc. Lond., vol. 17, 1885, pp. 4^-11; H. Lamb, Phil. Trans., vol. 203 A, 1904, pp. 1-42. It has also been suggested that the long waves are those distortional waves which suffer innumerable re- flections at the surface (sect. 158) and, as it were, creep round beneath the surface: Knott (2), pp. 256-257. IX] PROPAGATION OF EARTHQUAKE- WAVES 151 travelling along it are transmitted in the shortest possible time. In addition, Knott makes use of Turner's table for P and S given abo^'e (sect. 150), and thus any errors that may exist in this table must in so far affect the accuracy of his results. The forms of seventeen complete seismic rays have been de- termined, ten for the primary waves and seven for the secondary waves. As both AvaA'es are of the same general character, the former only are represented in Fig. 65, which gives a section of a hemisphere through the focus, 7 Fig. Oj. Forms of seismic rays. With the exception of the seismic ray directed to the centre of the earth, every ray is notably curved. All those (numbered 1-5) which emerge at distances of less than 60° from the focus are concave towards the surface throughout their whole course. Of those which emerge at greater distances, the central and deeper jiortion becomes first nearly straight, as in ray 6, and then even slightly convex towards the surface, as in ray 7. This ray emerges at an arcual distance of 73° from the focus. The slight convexity in its form shows that the velocity has begun to decrease as the depth increases. The change takes place at a depth equal to about three-tenths of the earth's radius. The remaining rays (8-10), however, are practically straight through- out a large part of their courses, and this shows that the velocity at depths somewhat greater than the above amount does not vary much with increasing depth. 152 PROPAGATION OF EARTHQUAKE- WAVES [ch. 157. While the forms of the primary and secondary seismic rays are approximately similar, there is, however, a difference to be noticed in the variations of the velocities of the two types of waves in the interior of the earth. These velocities are repre- sented by the curves in Fig. 66, in which de^^ths below the surface in thousands of kilometres are measured along the hori- zontal axis (the radius of the earth being 6378 kilometres) and f~ 12 10 8 ( • / 6 A X ^0 1 2 3 Fig. 66. Diagram representing the variation of the primary and secondary wave-velocities witli tlie deptli. velocities in kms. per sec. in the perpendicular direction*. It will be seen that both curves are practically straight to a depth of about 700 kms. ; from 700 to 800 kms., there is a shght bend, after which the curves become again nearly straight to a depth of about 1400 kms. for the secondar}^ wave and about 1600 kms. for the primary wave. At still greater depths, the velocity of * These velocities, it should be noted, are actual velocities in the direction of the wave-path. The figures in sect. 151 are those for the mean surface velocity, which is of course greater than the actual velocity. IX] PROPAGATION OF EARTHQUAKE- WAVES 153 each wave is nearly constant, being 12'8 kms. per sec. for the primary, and 6-85 kms. per sec. for the secondary, wave. It will be seen (sect. 161) that the results of this and previous sections have an important bearing on our knowledge of the nature of the earth's interior*. 158. Internal Reflection of Primary and Secondary Waves. "When the primary and secondary waves meet the surface of the earth, they are reflected and refracted (sect. 142). As the second medium is either water or airt, the refracted wave is F Fij^. 67. liiagram illustrating the internal reflection of earthquake-waves at the earth's surface. condensational and possesses but a small part of the energj' of the incident waves; the remainder is distributed between the reflected condensational and distortional waves. These may arrive at a giver station after one or more reflections, and, though each wave after reflection will be of diminished energy, it may combine or interfere with wa\es proceeding by other routes and thus either strengthen or weaken those waves at the observing station. It is clear that an incident condensational wave may travel with one reflection from the focus F to the station S (Fig. 67) * Knott (8). t There may also be rellec-tion and refraction at the lower surface of the outer heterogeneous crust of the earth. 154 PROPAGATION OF EARTHQUAKE-WAVES [ch. in one of two ways : (i) the condensational wave may be reflected at the mid -point A of the arc ^.S", and the reflected condensa- tional wave continue by a similar path to S; or (ii) the con- densational wave may be reflected at a point B between F and A and the reflected distortional wave may proceed along the deeper course BS. Again, an incident distortional Avave may be reflected at A, and the reflected distortional wave continue to S; or the distortional wave may travel along the path FC (similar to BS) and the reflected condensational wave reach S by the shallower path CS. Similar reflections may occur at the points A', B', C on the major arc FS, and, if the constitution of the interior of the earth permitted their j^assage, the three waves would ultimately arrive at S. Waves also reach a given station S after two, three or more reflections, with or without change of type, and, though greatly weakened by such reflections, may help to reinforce vibrations proceeding by more direct routes. If, however, the angle of incidence be very nearly 90°, we have seen (sect. 14.2) that the wave may proceed unchanged in type without very material loss of energy. Such a wave would suffer innumerable reflections and practicall}^ would creep round the surface in the same way as the sound-waves are known to creep round the dome of St Paul's cathedral*. 159. The question now arises whether such reflected waves can be recognised on seismograms. Let us consider the record of the Bonin Islands earthquake of Nov. 24, 1914, illustrated in Fig. 63, and confine ourselves to single reflections without change of type of the condensational and distortional waves. From the seismogram, it is found (sect. 152) that the arcual distance of the origin is 93° and that the time at the origin was 11 h. 53 m. 20 s. Now, the point of reflection would be at a distance of 46|° from the focus, for which P — is 523 1 seconds, so that the total time taken by the ])rimary wave to travel from the focus to the point of reflection and on to the station would be 523| x 2 seconds or 17 m. 27 s. Thus, the time at which the reflected jDrimary wave would reach the station would be 11 h. 53'm. 20 s. + 17 m. 27 s., or 12 h. 10 m. 47 s. * Rayleigh, Theory of Sound, vol. 2, 1878, pp. 115-116; Wiechert, pp. 103-11.5. IX] PROPAGATION OF EARTHQUAKE- WAVES 155 Again, for the epicentral distance of 46|-°, the value oi S — O is 934^ seconds, so that the time taken by the secondary wave in travelhng from the focus to the point of reflection and on to the station would be 934| x 2 seconds or 31 m. 9 s. Thus, the time at which the reflected secondary wave would reach the station would be 11 h. 53 ni. 20 s. + 31 m. 9 s., or 12 h. 24 m. 29 s. It will be seen, from Fig. 63, that the pronounced movements marked PR and SR occur almost exactly at these times, and it may fairly be concluded that they are the reflected waves for which we are seeking*. 160. Returns of the Long Waves. As the large waves diverge mainly in two dimensions, they retain a large part of their energy at great distances from the epi- centre, and it is therefore to be expected that they should cross a given station more than once. In strong-earthquakes, they are usually recognised twice, and, in a few earthquakes, they have been re- corded three times. In their different transits, thev are known as the [F, , ^. „„ t^. .,, . .. ' ~ 1' tig. 68. Diagram illustiatino; JFgand JFgWaves. If, in the diagram the returns of the long waves. (Fig. 68), the circle represent a section of the earth through the focus F and the station S, it is clear that the Wi waves are those which travel along the minor arc FAS; the W2 waves those Avhich traverse the major arc FBS'f, and the J1\ waves those which, after passing S the first time, continue their journey round the world, pass through the focus F, and traverse for the second time the minor arc FAS. If ti , <2 and /g be the observed times of passage of the initial * It is doubtful whether waves suffering a single reflection at the mid- point of the major arc FS (P'ig. r»7) could he detected. After reflection and so long a course, the primary waves would |)rol)ably be too weak to influence the record, and. in any case, they would be confused by the long waves among which they would arrive. As the secondary waves are not dis- tinguishable more than 120" from the origin, and as half the major arc to a station at less distance must be greater than 120% it follows that, after a single reflection at A, they would never reach the station. f The W^ waves are clearly shown at W^ in the record of the Kangra earthquake of 1905 (Fig. 61). 156 PROPAGATION OF EARTHQUAKE- WAVES [ch. undulations of the W^ , W^ and W^ waves, and A the length in kilometres of the minor arc FAS, the mean velocities of the initial W^ and TFg waves are, respectively, 40000 - 2A , 40000 , and r — kms. per sec. ; ^2 1 3 1 40,000 kms. being the circumference of the earth and t^ — t^ the time taken by the JFg waves to travel completely roiuid the world. The average velocity of the initial fFg waves is found by Omori to be 3-7 kms. per sec. That of the lf\ waves, as it depends on but a few observations, is somewhat uncertain. From four earthquakes in 1900 and 1902, Omori gives 3 h. 20 m. 46 s. as the mean value of t^— t^, and 3-4 kms. per sec. as that of the velocity of the initial W^ waves. Other estimates of this velocity are 3-53 kms. per sec. for the Messina earthquake of 1908 (Galitzin) and 3-45 kms. per sec. for the Californian earthquake of 1906 (Davison). It will be noticed that the estimates of the velocity of the JFg waves are less than those of the W^ waves. The explanation no doubt is that the W^ waves are the representatives of the more prominent and somewhat later undulations of the long- waves — an explanation supported by Omori's estimates of the mean periods of the third phase (short-period waves) of the principal portion, and of the IFg and W^ waves, namely, 20-4, 20-4 and 19-4 seconds*. Nature of the Earth's Interior 161. As soon as it was realised that the secondary waves consist mainly of distortional waves, which cannot be trans- mitted by liquids, it was evident that continued scismological observations would extend our knowledge of the nature of the earth's interior. Since then, the chief contributions have been made by Oldham, Wiechert and Knott f. The present section is practically confined to the recent results obtained by Knott. * F. Omori, Publ. Eq. Inv. Com., No. 13, 1903, pp. 119-124. •f Oldham (2), Wiechert and Knott (3). Oldham's investigations were based on a small number of earthquake-records. The conclusions at which he arrived are: (i) that the interior of the earth, beneath the outer hetero- geneous layer (about 30 kms. thick), consists of uniform material that can transmit both condensational and distortional waves, and that tliis material IX] PROPAGATION OF EARTHQUAKE- WAVES 157 (i) The outer heterogeneous crust is clearly thin compared with the radius of the earth. The primary and secondary waves are distinctly separated at a distance of about 10° from the epicentre. The waA'es which emerge at this distance ha\e pene- trated to a depth of about 100 kms. As they must have traversed a considerable course within the homogeneous interior for the sifting-out of the wa\es to be accomplished, Oldham concludes that the outer crust, consisting of rocks such as we know at the surface, may attain a thickness of 30 kms., or about o^x^th part of the earth's radius. (ii) Reneath this outer crust lies a thick and practically homogeneous layer, in which the primary and secondary -waves become separated. The seismic rays of both waves are on the whole concaA e towards the surface, showing that the velocities increase with the depth until that depth is about equal to three- tenths of the earth's radius. After this, the velocities become constant, and then slightly decline for greater depths. (iii) This elastic solid shell probably extends to a depth of about half the earth's radius, but, since the maximum velocity of the secondary wave is reached some 200 kms. before that of the primary wa\'c, it would seem that, at about the above depth, the rigidity begins to break down. (iv) The most significant fact is the loss of the secondary waves at a distance of 1 20° from the epicentre. The distortional waves which emerge at this distance have penetrated to a depth between one-half and six-tenths of the earth's radius. Knott concludes, therefore, that, in the neighbourhood of the latter depth, the clastic solid shell gives place to a non-rigid nucleus of measurable compressibility, capable of transmitting con- densational, but not distortional, waves. Detekmixatiox ok the Epicextre of a Distaxt Earthquake 162. Of the earthquakes which are recorded at distant stations, the majority originate under the sea, and some .of the remainder in countries that are inhabited by uncivilised races. To determine the epicentres of such earthquakes, several methods have been devised. They depend on the approximate constancy with which undtTfioi's no marked (•liaii<,'c in pliysical cliaracter to a dcptli of aljout six-tenths of tlic radius; and (ii) tliat tlie central fotir-tcntiis of tlie radius are occupied by matter possessing radically different physical properties. 158 PROPAGATION OF EARTHQUAKE- WAVES [cii. the long waves travel, on the known duration of the preliminary tremors or of the first series only at given distances, and on the direction of the initial movement of the preliminary tremors. 163. (i) Method depending on Time-Observations (Milne's first method). This method depends on the time of arrival of the large undulations at four or more places. Let the times at the four places A, B, C, D (Fig. 69) be respectively Z^, t^, t^, t^, of which ti is the earliest. Let V be the velocity of the long Fig. 69. Diagram illustrating the determination of the epicentre from the times of arrival of the long waves. waves. When these waves reach the station A, they will be at distances V {t^ — t^), V {t^ - tj), V {t^ - t-^) from the stations B, C, D. Thus, if circles be described with the stations as centres and the above distances as radii, the first large waves will form a circle which passes through A and touches each of the three circles. ThecentreOof this circle is the epicentre of the earthquake. If the times were known at three stations A,B,C, only, two circles might be drawn through A to touch the circles with B and C as centres. The jiositionofthe epicentre might thus be indeterminate. Milne, who used this method, assumed that the velocity of the large waves was approximately 3 kms. per sec. or 1-6 degrees IX] PROPAGATION OF EARTHQUAKE-WAVES 159 of arc per min. The circles (represented as on a plane in Fig. 69) were drawn on a slate globe, and the circle with centre was drawn by trial. The results obtained by this method are obviously approxi- mate. They depend on the accuracy of absolute time-determina- tions at the different stations, and these may occasionally err by as much as 1 minute*. 164. (ii) Method depending on the Duration of the First Pre- liminary Tremor (Zeissig's method). If the value of 6' — P be known for a single station, the epicentre must lie on the circle with the station as centre and the corresponding distance as radius; if the values oi S — P he known for two stations, the epicentre must coincide with one of the two points of intersection of two circles ; if the values oiS — P be known for three stations, the particular point is determined. In practice, of course, the three circles seldom, if ever, intersect in a point. The epicentre is then taken to be the centre of the small triangle formed by the intersections of the three circles. ^This method, which is the most frequently used of all, gives results of greater accuracy than the first method, as it depends on measurements of the duration of an interval and not on the absolute times. As a rule, P and S can be determined to within a few seconds; but sometimes the initial tremors are so small that they fail to affect the less sensitive instruments. There may also be some inaccuracy arising from the fact that the tables for S — P give their average values for a large number of earthquakes. If L denote the time of arrival of the long undulations, similar methods are based on the values of L — P and L — S aX, three given stations. The first long undulations are, however, less sharply defined than the first movements of P and S, and thus tables depending on the values of L give results that are of inferior value to those obtained by Zeissig's method. The values of L — S may, however, give useful results when the valu? of P is indeterminate t. * J. Milne, Rep. Brit. Ass., 1899, p. 38: 1900, j). 79. t li M denote the time at wliieli tlic niaxiinuin large undulations begin, Milne has used a similar method based on the values of M - P at three stations. O. Klotz has devised a graphieal form of Zeissig's method whieli gives the position of the epicentre with ease and aecuraey (Hull. Seis. Soc. Amer., vol. 1, 1911. pp. 143-148). 160 PROPAGATION OF EARTHQUAKE-WAVES [ch. ix 165. (iii) Method depending on the Direction and Epicentral Distance (Galitzin's method). Prince Gahtzin has shown that the position of the epicentre may be determined from observa- tions at a single station. From the value of S — P for the station, we know the distance A of the epicentre, and thus that the epicentre lies on a circle with the station as centre and A as radius. Again, the N.-S. and E.-\V. components of the first displacement (which is not complicated by the superposition of other waves) give the direction of the movement, and this is found tocoincideveryclosely withthatof the epicentre from the station. The epicentre must thus lie at one or other end of the diameter in this direction, the particular end being determined by the vertical comiionent seismograph, and in the direction of the downward movement. The remarks on the previous method, so far as regards the evaluation of S — P, apply to this method. As an exami^le of this method, Gahtzin takes the case of the Monastir earthquake of Feb. 18, 1911. The record of the Galitzin seismograph at Pulkowa (59° 46' N. lat., 30° 19' E. long.) gives A = 2260 kms. or 20° 19' of arc, in the direction S.22°53'W. That of the Galitzin seismograph at Eskdalemuir (55° 19'N. lat., 3° 12' W. long.) gives A = 2360 kms. or 21° 14' of arc, in the direction S. 55° 58' E. The former record locates the epicentre in 40-5° N. lat., 20-1° E. long., the latter in 40-6° N. lat., 20-3° E. long.* 166. (iv) Method depending on the Direction only (Galitzin and ^Valker^s method). The last method leads naturally to the simplest of all methods, that which determines the epicentre by the intersection of the lines of direction at two stations. It possesses several advantages over other methods, in being inde- pendent of all estimates of the time, of the exact determination of the beginning of the second phase, and of the empirical tables for S — P. Lastly, although only two stations are used, there is no ambiguity as to the position of the epicentre. Applied to the Monastir earthquake of Feb. 18, 1911, the azi- muths at Pulkowa and Eskdalemuir assign 40-3° N. lat., 20-4° E. long, as the position of the epicentre, a result which agrees very closely with those obtained for the same places by Galitzin's method t. * B. Galitzin, Vorlesungen uber Seismometrie (1914), pp. 401—427; Cotnpt. Rend. Acad. Sci. Paris, vol. 150, 1910, pp. 642-645, 816-819. t Nature, vol. 90, 1912, p. 3. CHAPTER X GEOGRAPHICAL DISTRIBUTION OF EARTHQUAKES 167. De Montcssus de Ballore lias proposed a rough classifica- tion of countries depending on the number and strength of the earthquakes which visit them. He divides them into : (i) seismic coimtries, in which earthquakes are freqvient and sometimes disastrous; (ii) 2)eneseismic countries, in which earthquakes are severe, but fall short of destructive power; and (iii) aseismic coimtries, in which earthquakes are feeble or rare or even com- pletely unknown. Japan may be taken as a type of the first class, Switzerland of the second, and Russia or Brazil of the third, though, even in Brazil, 50 earthquakes have been recorded since the year 15G0*. Some conception of the varying seismicity of different countries may be obtained from the following figures. In Great Britain, 250 earthquakes originated during the 21 years 1889-1909. On the other hand, 8331 earthquakes were recorded in Japan during the eight years 1885-1892, and 3187 in Greece during the six years 1893-1898. Thus, taking area into account, for every earthquake felt in Great Britain, there were 50 in Japan and 158 in Greece. 168. In constructing a seismic map of a country, either earthquake-frequency alone is represented, or frequency in com- bination with intensity. De Montcssus de Ballore remarks, how- ever, that in some cases frequency and intensity are closely related. In Japan, for instance, he finds that, when the average disturbed area during a given interval increases or decreases, so also does the number of earthquakes in the same interval. The relation does not by any means hold uniformly, for, as at Lisbon in 1755 and Charleston in 1886, a disastrous shock may occur at a place in which earthquakes are far from numerous; * The principal work on the 8, pp. 71-72. 166 GEOGRAPHICAL DISTRIBUTION [CH. Thus, the seismic regions lying along the two great circles include 91 per cent, of all known earthquakes, while the con- tinental areas (notwithstanding their much greater extent) con- tain only 9 per cent. / ; ^ ^^. tijb A distribution so remarkable must be capable of some ex- planation, and this de Montessus de Ballore finds to be that the zones enclosing the seismic regions coincide exactly with the geosynclinals of the secondary epoch, as outlined by Haug. x] OF EARTHQUAKES 167 Thus, the unstable bands of the earth are those in which sedi- ments of great thickness have been intensely folded, dislocated and elevated in Tertiary times when the principal existing mountain-chains were formed; while the stable portions of the earth are those connected with the tabular architecture of the great continental regions*. 177. Very different in the materials employed, but not less interesting, are Milne's seismic maps of the world, which he issued annually in the reports of the Seismological Committee of the British Association. In dealing only with earthquakes felt over an area not less than that of Europe and Asia com- bined, Milne has greatly simplified the problem of seismic dis- tribution. His maps, one of which is reproduced in Fig. 71, show the regions of the earth in which the epicentres of the ten years 1899-1908 are situatedt. Their only defect is not one of prin- ciple, but is simply due to the brevity of the period considered; for, in the history of terrestrial change, ten years is a length of time almost negligible. As will be seen in a later section (sect. 183), seismic activity on a great scale is svibject to con- siderable migration. Except for minor shocks, a region once struck by a violent earthquake tends to remain quiescent for a prolonged interval. In this respect, Milne's maps arc less in- structive than de Montessus de Ballore's. On the other hand, in their completeness, they possess a compensating merit, for no great earthquake, wherever it may occur, can now escape de- tection and registration. In the map given in Fig. 71, twelve districts are represented, roughly bounded by oval curves^. It will be seen that four of them {G, H, I and J) are entirely oceanic, one of them {K) terrestrial, while six {A, B, C\ D, E, F) are partly oceanic and partly terrestrial. As the district C (including the West Indies and the Caribbean Sea) belongs technically to the Pacific, it will be seen that these six districts cling to the east and west margins of that ocean. * De Montessus de Ballore, pp. 2;j-2f». t The method of determining tlie position of the epicentre is cxphiined in sect. 103. X This map is reproduced from the Reiinrl (if the Seismological Committee for 1 <)()». In the latest re|)orts, tlic majjs arc more (U-tailcd and less suited for the |)urpose of this chapter. They should, however, he consulted.' 168 GEOGRAPHICAL DISTRIBUTION [ch. The numbers of earthquakes originating during the years 1899-1908 in the different districts are as follows: A 40, B 55, C 30, D 28, E 133, F 175, G 26, H 35, 1 5, J 5, and K 141. Thus, the four oceanic regions contain 11 per cent, of the total number of earthquakes, the great terrestrial region (K) 21 per cent., and the six regions {A-F) bordering the Pacific Ocean 68 per cent. Thus, roughly, of every ten great earthquakes, seven occur near the margins of the Pacific, tAvo in the great land- region, and one in the oceanic regions. At the present time, the great unstable region of the earth lies along the west margin of the Pacific, including Japan, the Philippine Islands and the Malay Archipelago. Of the earth- quakes in the six Pacific regions, the three along the east margin contain 33 per cent, of the total number, while those along the west margin contain 67 per cent. Thus, approximately, for every two great earthquakes felt along the east margin, there are five along the western side. Laws of Seismic Distribution 178. Connexion with the Gradient. The most important law of seismic distribution is that earthquakes are as a rule strongest and most frequent in those portions of the earth in which the average slope of the grovmd is greatest. On a large scale, this fact is evident from the maps of the whole world described in sects. 175-177. The bands of seismic activity depicted by Mallet generally follow the lines of eleva- tion which divide the great basins of the earth's surface and shun the central portions of those basins. Again, the margins of the Atlantic slope gently towards the central basin, except in the Gulf of Mexico and the Caribbean Sea; those of the Pacific, and especially along its western border, dip steeply towards the great known depths of ocean. Now, the Atlantic earthquakes amount to 7 per cent, of those registered by Milne during the years 1899-1908. In contrast with these figures, the Pacific earthquakes number 71 per cent, of the whole, those on the west side 48 per cent, and those on the east side 23 per cent. Again, the great terrestrial region {K) includes four sub- regiojis, namely, the Alpine, the Balkan, the Caucasian and the X] OF EARTHQUAKES 169 Himalayan. It is in the latter, which includes 63 per cent, of the earthquakes belonging to the whole region, that the most pronounced foldings occur, and that, in distances of 100 miles, gradients of 1 in 44 are to be foiuid. 179. The same law holds true for smaller areas than those considered in the last section. De Montessus de Ballore, who in- cluded in his siu'vey earthquakes of all degrees of strength, states the law in the following precise manner: "in a general way, we may say that, of two contiguous regions — for example, the two slopes of a valley, the two flanks of a mountain-chain, plains and neighbouring heights, etc. — the more unstable is that which presents the greater average slope." In the same way, Milne, in his study of the Japanese earthquakes of 1885-1 892, concludes that "where there is the greatest bending, it is there that sudden yielding is the most frequent." The epicentre of the Sanriku earthquake of June 15, 1896, for instance, was at a depth of 4000 fathoms, exactly at the bottom of the western slope of the well-known Tuscarora Deep (Fig. 72). The earthquakes of the Japanese Empire are worthy of study in greater detail. These islands are arranged in the form of a festoon with its convexity facing the Pacific Ocean, and, as in similar groups of islands and in mountain-chains of the same form, the convex side of the festoon slopes more steeply than the concave side. As already indicated, the earthquakes of this country follow a law which is general in such cases; they are more numerous and more violent on the steeply sloping convex side than on the other. The dotted lines in Fig. 72 represent contour-lines of the sea- bed (for each thousand metres) on both sides of the country. The Japan Sea is shallow, its greatest depth being only 3000 metres (1646 fathoms). The gradient of the sea-bed varies from 1 in 67 to 1 in 110; the average gradient to a depth of 1000 metres for the coast of Hokkaido is 1 in 220. On the other hand, the Pacific Ocean is vcrydcej). The extraordhiary basin, called the Tuscarora Deep, reaches a depth of 8000 metres (4376 fathoms) at distances of 110 to 240 miles from the coasts. The gradient is unusually steep, being 1 in 27 off the coast of Xemuro. 1 in 30 oif the north-east coast of the main island, and 1 in 16 (to a depth of 3000 metres) off the south-cast coast of Kazusa and Awa. 170 GEOGRAPHICAL DISTRIBUTION [CH. The origins of 221 destructive Japanese earthquakes from the fifth century to the present time have been investigated by Omori. Of the total number, 114 originated inland, 47 under the Pacific Ocean, 17 under the Sea of Japan, 2 under inland seas, while the epicentres of 41 earthquakes are unknown. Ten of these earthquakes were very violent, and, while three of them Fig. 72. Map of seispiic regions in Japan. occurred in central Japan, seven originated off the south-east coast, each of the latter being accompanied by seismic sea- waves. Again, on the Pacific coasts, there were 23 great sea- waves dviring the period mentioned, while on the Japan Sea coast there were only five small sea-waves. Similar results follow from Omori's examination of recent strong Japanese earthquakes. From 1885, when the systematic x] OF EARTHQUAKES 171 observation of earthquakes was begun, to 1905 (that is, in 21 years), 257 earthquakes originated in or around Japan, some of which were destructive or semi-destruetive, Avhile the rest were strong or moderate shocks, each having disturbed a land- area of more than about 25,000 sq. miles. The principal regions into which the epicentres of these strong shocks are grouped are represented by the broken lines in Fig. 72, which is reproduced from Omori's map. During the 21 years re- ferred to, 138 earthquakes originated in the region A, 7 in the region C, 2 in F, 4 in G, 3 in H,12 in ^,16 in M, 12 in A' and 23 in P. The most active region at jDresent is thus that marked A, stretching off the east coasts of Hokkaido and Main Island, the number of earthquakes which occurred in it being rather more than half the total number in Japan. The zone a, indicated by a broken line, represents approximately the eijicentre of the great earthquake of Oct. 28, 1707 (the greatest of all Japanese earthquakes), and the two great shocks of Dec. 23 and 24, 1854. The zone a evidently connects the regions A and C, the whole forming the principal sub-oceanic earthquake-region of Japan, and called by Omori the External Seismic Zone. The transference of activity, from the zone a in former times to the region A in the present day, is worthy of notice. On the Japan Sea side, there are three regions. F, G and H, including between them not more than nine earthquakes. The most important region is F, which in past times has produced some violent shocks. The zones, b and c, represent approximately the positions of the epicentres of the great earthquakes of Dec. 7, 1833, and March 14, 1872. These two zones, with the regions F, G and H, thus form a continuous region bordering the concave side of the Japanese islands, and called by Omori the Inner Seismic Zone. During the 21 years, 1885-1905, there were thus nine earth- quakes in the inner, and 145 earthquakes in the outer, seismic zone. In other words, for every strong earthquake originating on the concave Japan Sea side of the islands, there were 16 on the convex Pacific side*. * De Montessus de Ballorc, |)|). 18-10; J. Milne, Seis. Journ., vol. 4, 1895, pp. xv-xvi, sind lief). Brit. Ass., 1897, pp. 29-30; F. Omori, Hull. Eq. Jnv. Com., vol. 1, 1907, pp. 114r-123. 172 GEOGRAPHICAL DISTRIBUTION [CH. 180. The earthquakes of the Japanese Empire have been con- sidered in some detail, for they have been studied more carefully than those of any other country. In other groups of islands arranged in the festoon form, such as Sumatra and Java and the Aleutian Islands, or in mountain-chains like the Himalayas and those of Alaska, the same law of distribution prevails. The steep convex side is visited by more frequent and more violent earthquakes than the gently-sloping concave side. Fig. 73. Distribution of earthquake-zones in Calabria. 181. Earthquakes and Secular Changes of Elevation. A jDOSsible relation, on which further information is required, is one pointed out by Milne regarding the occurrence in Japan of districts which are known to be undergoing a slow process of elevation. Of the 15 seismic districts into which he divides the whole coimtry, ten show evidences of very recent elevation, and in five of them earthquakes are extremely frequent. There are, however, two exceptions to this rule, Avhich he points out. One seismic district is, so far as known, free from any move- ment, either of depression or elevation; while another section X] OF EARTHQUAKES 173 ^iuikamachi Nagano /O o;' Matsumoto of the coast, which is known to be rising, is comparatively free from earthquakes*. * MiGRATIOXS OF SeISMIC ACTIVITY 182. Small Migrations of the Epicentre. Detailed mapping of successive earthquakes in any seismic region shows that the epicentres are subject to continual migration. Two examples of this migration are given in other chapters, the after-shocks of the Inverness earthquake of 1901 being considered in sect. 240, and those of the Mino-Owari earthquake of 1891 in sect. 215. The migration of earthquake- centres in another district, that of southern Calabria, is represented in Fig. 73. The shaded areas are those most strongly shaken in different earthquakes, and therefore must closely surround the corresponding- epicentres. The areas shaded with horizontal lines belong to the re- markable series of Calabrian earth- quakes in 1783 and the following years. Those shaded with vertical lines relate to the earthquakes of 1905, while those shaded obliquely are the central areas of other earthquakes. In 1783, the first great earthquake occurred on Feb. 5 in the Palmi zone, the second a few hours afterwards in the Scilla zone. Two days later, the third great earthquake of the series took place in the Monteleonc zone, followed in two hours by the fourth in the Messina and Scilla zones. On Mar. 5, the fifth great earthquake occurred in the Monteleonc zone, and, on Mar. 28, a sixth, almost, if not quite, as strong as the first of the series, in the Girifaleo zone. Sometimes, the different areas are shaken simultaneously, or perhaps in very rapid succession. In the great earthquake of 1905, this was the case in five zones, namely, those of Palmi, Monteleonc, Nicastro, Cosenza and * Seis. Journ., vol. 4, 1805, p. xvi. Fig. 74. Distribution of earthquakes in central Japan. 174 GEOGRAPHICAL DISTRIBUTION [ch. Bisagnano. At other times, single areas only are affected, such as that of Nicastro in 1638, Montclebne in 1659, Bisagnano in 1836, Cosenza in 1854, Palmi in 1894, and Messina in 1908. The district represented in Fig. 74 is the Shinano-gawa earth- quake-zone in Japan (M, in Fig. 72). The small ovals indicate the approximate positions of the epicentres, the numerals attached to them showing the order of their occurrence. All were severe local shocks, strong enough to produce cracks in the ground and to cause some injury to buildings. The first earthquake of the series occurred on July 23, 1886, and the successive shocks on July 22, 1887, Jan. 7, 1890, Jan. 17, 1897 and Jan. 22, 1899*. 183. Large Migrations of the Epicentre. The transference of seismic activity is equally characteristic of larger areas. For instance, in the Iberian peninsula, the western portion was chiefly affected in the eighteenth century, and the southern districts in the nineteenth. In the former century, Portugal was frequently shaken, the great Lisbon earthquake tending to close the series in 1755. In the next century, Portugal was almost unshaken, Avhile destructive earthquakes occurred in Almeria in 1804, 1860 and 1863, in Mureia in 1828-29 and 1864, and in Andalusia in 1884. On a still larger scale is the migration of seismic activity along the west coast of the American continent from 1899 to 1912. On Sep. 4 and 11, 1899, and Oct. 9, 1900, there were three great earthquakes in Alaska; on Jan. 20, 1900, and Apr. 19 and Sep. 23, 1902, there were destructive shocks in Mexico, Guatemala and other parts of Central America; on Jan. 31, 1906, a seventh in Panama and along the west coast of Colombia and Ecuador, followed on Apr. 18 by the Californian earthquake, and on Aug. 17 of the same year, by the great earthquake in Valparaiso. On Apr. 15, 1907, and Nov. 19, 1912, disastrous earthquakes took place in Mexico. By the epicentres of these eleven earth- quakes, the western coast of America may almost be said to have been outlined from the extreme north-Avest to as far south as Chilif . * M. Baratta, Sopra le zone sismologicamente pericolose delle Calabrie, Voghera (no date); F. Omori, Bu«. Eq. Inv. Com., vol. 1, 1907, pp. 138-141. t Bull. Eq. Inv. Com., vol. 1, 1907, pp. 21-23. X] OF EARTHQUAKES 175 Distribution of Submarine Earthquakes 184. Submarine earthquakes belong to three classes : (i) those which originate so near land that they are felt along- the ad- joining coasts; (ii) those which are felt on passing ships; and (iii) those which are strong enough to be registered at some or all seismological stations. The distribution of earthquakes in the third group is considered in sect. 177, from which it is clear that a large number of great earthquakes are certainly of sub- marine origin. How frequent are earthquakes of the first class is evident from the observations described above (sect. 179), In a seismic country like Japan, it would seem that at least half the earthquakes originate beneath the ocean. 33 30 25 20 15 : .•••■. '"■■■-:5.SH. K)2- 4():j; .J. .Millie, Hep. lirit. Ass., 1»11, pp. 04U-7K). 12—2 180 FREQUENCY AND PERIODICITY [ch. which follow them, he finds the average intensity of the groups for each of such intervals. The interval between the centres of successive groups usually varies from 15 to 50 days, and is roughly proportional to the intensity of the earlier group, a group of great intensity being followed by a long period, and one of low intensity by a short period, of quiescence *. 189. Double and Multiple Earthquakes. In addition to this tendency to clustering, great earthquakes, as Milne has shown, often occur in pairs and even in triplets in widely separated districts. A typical example is that of Apr. 19, 1902. At 2.21 p.m., a great earthquake destroyed many towns and villages in Guate- mala. The seismographs at Capetown, Calcutta, Bombay, and other places record its vibrations, but they also reveal the waves of a second earthquake which, according to Milne, originated at about 2.34 p.m. in the Indian Ocean in 35° S, lat., 60° E. long., approximately. The difference in time between the two earth- quakes is thus about 14 minutes and the distance between their epicentres is 146°, the time required by the first preliminary tremors to traverse this arc being about 17j minutes. Again, on Aug. 17, 1906, a great earthquake occurred at h. 8 m. or Oh. 11 m. p.m., its epicentre lying in 31° N. lat., 168° E. long. On the same day, at h. 41 m. p.m., Valparaiso and the neighbouring towns were ruined. The interval between the two earthquakes was thus about 33 minutes. To traverse the distance of 122° between the two epicentres, the first series of preliminary tremors would require nearly 16 minutes and the second series about 29 minutes. During the years 1899-1906, there were, as Milne points out, 15 cases of great double earthquakes and 2 of triple earthquakes, the intervals between them varying from 2 to 106 minutes, with an average of 27 minutes. As the average interval between successive great earthquakes amounts to nearly a week, it is thus difficult to suppose that, in these pairs, the occurrence of the second earthquake was independent of that of the first |. * F. Omori, Journ. Coll. Sci., Imp. Univ. Tokyo, vol. 11, 1899, pp. 410- 412; J. Milne, Rep. Brit. Ass., 1910, pp. 54-55; 1912, pp. 92-94. t J. Milne, Rep. Brit. Ass., 1911, pp. 32-35. Milne suggests that every great earthquake causes a relief of seismic strain throughout the world, and that the disturbance in the second focus results from the arrival of the waves from the first. In about one-half the pairs, however, the XI] OF EARTHQUAKES 181 190. Synchronous Variations of Frequency in Different Dis- tricts. Milne has shown that, in some widely separated regions, the variations of frequency may be synchronous or nearly so. Taking the East Pacific districts A, B and D (sect. 177) and the West Indies C in one group, and the West Pacific districts E and F, and the Himalayas K in the other, the following table gives the totals for the vears 1899 to 1907: 1899 1900 1901 1902 1903 1904 1905 1906 1907 Total E. Pacific 31 20 18 18 13 6 13 14 W.Pacific 30 16 27 45 41 28 40 53 38 133 318 Thus, the greater seismic activity has been manifested on the Asiatic side of the Pacific Ocean and especially in the Malay Archipelago, while the west side of South America has been least frequently disturbed. Also, since 1902, the totals for the two great regions rise and fall together in successive years. The synchronism, however, appears to extend to still more widely separated regions, such as the Italian peninsula and Sicily; Japan, Formosa and the Philippines; North, Central and South America; and China. Considering destructive earth- quakes only for the 200 years 1700-1899, Milne divides the whole period into six intervals, the first from 1700 to 1734, and each of the others of 33 years each. Any given year is considered to be one of activity or quiescence if the number of destructive earthquakes in any one region be above or below the average for the 33 years in which it occurs. Now, all four districts have shown abnormal seismic activity in 12 years and comparative quiescence in 15 years; three districts have been unusually active and one quiescent in 46 years, and three quiescent and one active in 58 years. Thus, in 131 years out of 200, or roughly in two out of three, seismic activity or quiescence has generally jirevailed in these four widely separated regions of the world. It would seem from this that the seismic activity of a district does not depend only on local conditions. iMit may be partly interval between tlie eartli(|iiakes is actually less than tiie time required by the first preliminary tremors to traverse the intervening distanee. Is it possible that these double eartlupiakes are analo!). 2. (2). On the diurnal i)eri()dieity of earthquakes. Phil. Mag., vol. 42, 189H, pp. 4«8-47(). ;{. Knott, C. G. Eartlupiake t're(|ueii(y. Trans. Svis. Soc. Japan, vol. !), 188(>, pp. 1-20. 4. Schuster, A. On lunar and solar periodicities of eartlKiuakes. Proc. Half. Soc, vol. 1 \ \ t / / \ ^ X ^ •s y \ y / \ / \ \, / -^ :> \ \ / N ^ > ■^ •> s / XI] OF EARTHQUAKES 185 « and diurnal periods, for the numbers which the curves represent may be due to co-existent periods of various lengths. There may. for instance, be a semi-diurnal, as well as a diurnal, seismic period, and the curves in Fig. 77 may be the resultant of these two, and possibly of other, periods. 194. Let us suppose that the numbers of earthquakes during successive hours of the day are the sums of corresponding numbers in the following series : 4, 13, 29, 50, 73, 100, 127, 150, 171, 187, 196, 200, 196, 187, 171, 150, 127, 100, 73, 50, 29, 13, 4, 0, and 6, 25, 50, 75, 93. 100, 93, 75, 50, 25, 6, 0, 6, 25, 50, 75, 93, 100, 93,. 75. 50, 25, 6, 0. If the numbers in the first series be plotted, the curve so ob- tained (represented by the broken line in Fig. 78) is the curve 0-1 2-3 4-5 6-7 8-9 10-11 12-1 2-3 4-5 6-7 8-9 IQ-H Fig. 78. Harmonic curves and their compound. of sines, the period being one day, and the maximum epoch being at the middle of the hour 11-12 a.m., or 11.30 a.m. If the numbers in the second series be plotted, the curve so obtained (represented by the dotted line in Fig. 78) is again the curve of sines, but its period is 12 hours, and the maximimi epochs occur at 5.30 a.m. and p.m. Adding together the corresponding ntunbers in the above two scries, we obtain the following series: 10. 38. 79. 125, 160. 200. 220. 225, 221, 212, 202, 200, 202, 212, 221, 225, 220, 200, 166, 125, 79, 38, 10, 0. The curve representing these numbers is the continuous curve in Fig. 78. in which there are two maxima, at 7.30 a.m. and 3.30 p.m., with a slightly marked minimum at 11.30 a.m. and a j^ronouiieed ininiiuuin at 11 ..30 ]).m. 195. Method of Analysis. The variation in earthquake-fre- quency may be irregular, as shown by the curves in Fig. 77, or it mav rise and decline graduallv. as indicated by the con- 186 FREQUENCY AND PERIODICITY [ch. tiniious curve in Fig. 78. But, whatever the cur^•e of variation may be, it is always possible to represent it by adding together the ordinates of a series of sine-curves with periods of re- spectively 1, I, J, ^, etc., that of the period investigated, and the object of harmonic analysis is to sift out each of these periods from the others and to ascertain its amplitude and maximiun epoch. In determining these elements of the various seismic periods, the ordinary method of harmonic analysis may be used, and the results may be calculated to any degree of accuracy desired. It is important, however, to avoid giving an appearance of pre cision where no real precision is attainable. Only those cata- logues which are based on instrumental records can lay any claim to completeness, and but few of these cover a period of years sufficient to neutralise the tendency of earthquakes to occur in groups. Thus, imtil our materials approach complete- ness, the somewhat rough form of harmonic analysis known as the method of overlapping means gives results which are quite as acciu'ate as oiu* materials warrant. This method will now be described*. 196. For the annual period, the numbers of shocks in each month are counted, and these numbers are taken to represent the rate of occurrence at the middle of the month. Six-monthly means are then calculated for each month, the mean for January being that of the six months from November to April inclusive. The middle of that period being at the end of Janviary, it follows that the mean so obtained corresponds to the end of the month. The effect of taking six-monthly means is to eliminate the semi- annual period, if there be one, and to diminish or eliminate all other minor periods, and thus to extricate the annual period almost entirely free from others which are not being examined. To compare the results for any district with those of other districts, each six-monthly mean shoidd be divided by the a^xrage of all twelve means, that is, by the average monthly number of earthqviakes. The method of overlapping means, how- ever, reduces each such mean in the ratio 1-589 to 1, and the difference between each reduced six-monthly mean and unity must therefore be augmented in this ratio. * Knott, pp. 8-10, 20; Davison (1), pp. 1108-1115. XI] OF EARTHQUAKES 187 For the semi-annual ])eriod, the method is the same, except that the numbers of shocks in successive half-months are counted, and those for the first halves of January and July are added together, and so on. The same methods are followed for the diiu-nal and semi- diurnal periods, except that, for the former, it is necessary to take twelve-hourly means. To obtain that for the hour 0-1 a.m., the mean of the 12 hours from 7-8 p.m. to 5-6 a.m. is calculated, and is taken to correspond to the middle of the interval, namely, to 1 a.m. The augmenting factor in this case is 1-522. 197. Let us apply this method to the third series of numbers in sect. 194. Each number in this series is the sum of the corre- sponding numbers in the first and second series. The former have a maximum at 11.30 a.m., the latter have maxima at 5.30 a.m. and p.m. The amplitudes of the two series of nimibcrs are respectively 100 and 50, the ratios of which to the average (150) of the third series of numbers are \^^ and j^^j or -67 and -33. Adding together the numbers for the 12 hours from 7 p.m. to 7 a.m., 8 p.m. to 8 a.m., and so on, we obtain the following twelve-hourly means (each nuiltiplied by 12) corresponding to the middle of each 12 hom-s, that is, to 1 a.m., 2 a.m., etc.: 1090, 1190, 1332, 1506, 1698, 1898, 2090, 2264, 2406, 2506, 2560, 2560, 2506, 2406, 2264, 2090, 1898, 1698, 1506, 1332, 1190, 1090, 1036, 1036. Dividing each of these numbers by their average, 1798, and multiplying the difference between each and unity by the aug- menting factor for twelve-hour intervals, 1-522, we obtain the following numbers : -.37, -46, -59, -73, -91. 1()9, 1-27, 1-41, 1-54, 1-63, 1-66, 1-66, 1-63, 1-54, 1-41, 1-27, 109, -91, -73, -59, -46, -37, -33, -33. These numbers represent the ratio of the numbers of earth- quakes during each hour to the average hourly number, so far as they arc due to the diurnal period. As the highest mnnbcr, 1-66, corresponds to 11 a.m. and noon, we coucliule that the maximum of the diurnal period occurs at 1 1 .30 a.m. and that its amplitude is -00, or slightly more. For the semi-diurnal period, the numbers for corresjionding hours, a.m. and ]).jn., are added together, namely, 212, 250, .300, 350, 386, 400. 386, 350, .300, 250, 212, 200. 188 FREQUENCY AND PERIODICITY [ch. Adding together the numbers for the 6 hours from 10 to 4, 11 to 5 and so on, we obtain the following six-hourly means (each multiplied by 6), corresponding to the middle of each 6 hours, that is, to 1, 2, etc., a.m. and p.m.: 1524, 1698, 1898, 2072, 2172, 2172, 2072. 1898, 1698, 1524, 1424, 1424. Dividing each number by the average of all, 1798, and multi- plying the differences between each and unity by the aug- menting factor for six-hour intervals, 1-589, we obtain the following numbers: •76, -92, 1-08, 1-24, 1-33, 1-33, 1-24, 108, -92, -76, -67, -67. These numbers represent the ratio of the number of earthquakes during each hour, so far as they are due to the semi-diurnal period. As the highest number, 1-33. corresponds to 5 and 6 a.m., we conclude that the maxima of the semi-diurnal period occur at 5.30 a.m. and p.m. and that the amplitude of the period is -33 or slightly more. For the eight-hour period, the numbers for 0-1 a.m., 8-9 a.m. and 4-5 p.m. are added together, and so on. We thus obtain the following numbers: 451, 450, 447, 450, 447, 450, 451, 450. For the six-hour period, the nvimbers for 0-1 a.m., 6-7 a.m., 0-1 p.m. and 6-7 p.m. are added together, and so on, and we obtain the numbers : 598, 600, 600, 600, 598, 600. The numbers in each series are practically equal, and this shows that both the eight-hour period and the six-hour period are absent from the variation in frequency considered. The analysis has thus sifted out with accurac}^ the required component periods. 198. It is important that the catalogues which are subject to analysis should be as complete as possible and should cover a long series of years. If the total mnnber of earthquakes in- cluded should be small, the analysis of the figures may provide a period which does not really exist. Schuster has shown that, if the earthquakes should occur at random, harmonic analysis may give rise to an apparent seismic period of amplitude -\/(7r/w), where n is the number of earthquakes. Moreover, earth- quakes occur, not at random, but in groups. Thus, in any isolated XI] OF EARTHQUAKES 189 record, unless the amplitude be much in excess of the above amount or expectancy, the period indicated by the analysis cannot be regarded as established. If, however, the maximum epochs of a particular period should agree approximately in neighbouring districts or for the same district in different in- tervals of time, the existence of the period may be regarded as probable even if the calculated amplitude were to fall below the expectancy. 1-6 " - 1-5 -•. - 1-4 - ^^\ 1-3 \ \ 1 -2 - ^ ■. \ / ■ / "~^ ' \ / ■ / 11 10 " • ^ \ '■ , ^ \ , '. / / / - J F M VA M J J A .S' / N D ■9 - • ^ \ • ^ \ ■. \ \ , ■ J / - •8 ■■, \ \ ■ '^ \ '• \ V • ~ •7 " / ^^ "" •6 - '•. ""' - ■5 - ■4 _ 1 1 1 1 1 1 1 1 1 1 Fif. 80 differs slii^htly, owing to tlie mode of treating the statisties, from that hiid down hy Omori. Three small detaehed areas with a simniier maximum are also omitted, owing to the cxelusion of the records from the twelve stations mentioned above. t F. Omori, Publ. Eq. Itiv. Com., No. 8, lUO'i, pp. 1-91; No. 18, 1904, pp. 23-2(}; Hull. Kq. Inv. Com., vol. 2, 1908, pp. 3;>-50; vol. 5, 1913, pp. ;J9-8G. D..M.S. 13 194 FREQUENCY AND PERIODICITY [CH. is less than 600 sq. miles, between 600 and 6000, or more than 6000 sq. miles, with the following results: * No. of earth- quakes Expect- ancy Amiual Period Max. epoch Ampl. SHght Moderate Strong 2256 567 176 •04 •07 •13 Middle of Oct. Middle of Mar. End of Mar. •14 •17 •17 Three different catalogues of the earthquakes of Zante give similar results, the third catalogue differing from the second merely in including a larger number of very slight shocks *. Catalogue No. of earth- quakes Expect- ancy Annual Period Max. epoch Ampl. Schmidt and Fuchs Barbiani (1) „ (2) 246 1326 1663 •11 •05 •04 End of Dec. End of Aug. •29 •10 •29 203. Diurnal Seismic Period. In the analysis of the diurnal periodicity of earthquakes, it is essential that the records used should be entirely instrumental (sect. 193). Such records are few in number and not of long duration. The method of har- monic analysis has been applied to those of Tokyo, Japan, Manila, and various Italian observatories. The following table gives the results obtained by the method of overlapping means : District Tokyo, whole year „ winter „ summer Oita Japan Manila Italy No. of earth- Expec- quakes tancy 2539 •04 1290 •05 1249 •05 237 •12 1175 •05 208 •12 8177 •02 Diurnal Max. epoch 11 a.m. 12 noon 11 a.m. 11 a.m. 12 noon 11 a.m. 1 p.m. Ampl. •08 •06 •10 •39 •11 •30 •32 Semi-diurnal Max. epoch a.m. &p.m. 9 9 9 8 irregular 3 11 Ampl. •10 •13 •06 •12 •05 •19 •11 * Davison (1), pp. 1116-1120. According to Omori (Bull. Eq. Inv. Com,, vol. 2, 1908, pp. 17-20), the same reversal characterises the weak and strong earthquakes of Tokyo and Kyoto. XI] OF EARTHQUAKES 195 The diurnal period is thus less clearly marked than the annual period. In most cases, the amplitude does not greatly exceed the exiDcctancy. The effect of the tendency to occur in groups is, however, shghter than in the annual period ; and there is also a close agreeent in the maximum epoch, not only in different districts, but in the same district during the winter and summer months. So far as we may judge from the records at our dis- posal, there seems to be a diurnal period with its maximum between 11 a.m. and 1 p.m. The semi-diurnal period is still less pronounced, the average amplitude being -10 instead of '19. The maximum epoch is variable, except in Japan, where it occurs about 9 a.m. and p.m. 204. After a great earthquake, the crust is in a peculiarly sensitive condition^ and fluctuations in frequency should then be strongly marked. Instrumental records have been obtained of the after-shocks of the Mino-Owari earthquake of Oct. 28, 1891, at Gifu and Nagoya, and of the Hokkaido earthquake of Mar. 22, 1894. The results arc given in the following table: Diurnal Semi-diurnal No. of earth- Expec- District Max, epoch a.m.&p.m. quakes tancy Max. epoch Ampl. Ampl. Gifu: Oct. 29-Nov. 10, 1891 1257 •05 12 noon •20 6 •11 Nov.ll,1891-Dec.31,1899 2856 •03 0| a.m. •13 8^ •06 Nagoya : Oct. 29-Nov. 10, 1891 572 ■07 3 a.m. •35 2 •19 Nov. 11, 1891-Dec. 31,1899 1282 •05 about midn. •14 3 •11 Nemuro : Mar. 2:5-3 1, 1894 345 •10 H a.m. ■S6 7 •25 Apr. 1, l«»+-l)ec. 31. 1899 646 •07 0.V p.m. •16 5 •13 The diurnal variation in the two Xagoya records is repre- sented in Fig. 81, in which the continuous line indicates the variation during the earlier period and the broken line that during the later. Thus, in every case, the amplitude of the diurnal period is about three times, or more than three times, the expectancy. We may therefore conclude that: (i) after-shocks of great earth- quakes are governed by a diurnal fluctuation in frequency, the 13—2 196 FREQUENCY AND PERIODICITY [CH. maximum occurring at, or shortly after, midnight, and (ii) the diurnal period is more marked during the week or ten days following the principal earthquake than afterwards. The semi- diurnal period is less pronounced than the diurnal period, and is more marked during the week following the principal earth- quake than afterwards; the maximum epochs are variable. As the diurnal barometric period has a maximum epoch a few hours after midnight, it is possible that the diurnal variation in the frequency of after-shocks may be connected with the diurnal variation in barometric pressure*. V4 1-3 1-2 1 1 1-0 •9 •8 •7 Fig. 81. Diurnal periodicity of the after-shocks of the Mino-Owari earth- quake of Oct. 28, 1891, at Nagoya, (i) Oct. 29-Nov. 10, 1891, (ii) Nov. 11, 1891-Dec. 31, 1899. J ! \ I \ l\ I •• I I 1 I I I \ 1 I I LXI H i L 2-3 6-7 ••. 10-11 2-3 6-7/ 10-11 205. Relation between Variations of Latitude and the Fre- quency of Earthquakes. The recent discovery of the variations of the Pole from its mean position has led Milne to make an interesting comparison between the frequency of great earth- * Davison (2), pp. 465—476. Omori has examined a large number of seismic records from different observatories in Japan. Most of them are non-instrumental for part of the time. The method of analysis used by him, moreover, is practically that of taking two-hourly overlapping means — a form of smoothing which is not well adapted for the purpose in view. By combining in one diagram, the diurnal variations in earthquake-fre- quency and barometric pressure throughout the day, Omori is led to con- sider the former variation as a result of the latter (Publ. Eq. Itiv. Com., No. 8, 1902, pp. 53-94). XI] OF EARTHQUAKES 197 quakes and the changes in direction (or deflections) of the move- ments of the Pole. The comparison is made for the years 1892 to 1904. Dividing the year into ten intervals of 36^ days each, Milne finds the numbers of great earthquakes in the intervals in which deflection occurs and in those immediately before and after them. During the years considered, there were 23 deflection- periods, and in 18 of these the number of earthquakes in the deflection-period was greater than in either adjoining period. The total number of earthquakes in the deflection-periods was 287, during those preceding them 167, and during those following them 217 — numbers which are as 100 : 58 : 76. Milne thus con- cludes that great earthquakes are frequent about the times when changes occur in the direction of the polar displacements and especially when those rates of change are rapid. That the deflections of the polar movements govern the fre- quency of the stronger earthquakes only seems clear from a comparison which Omori has made for the interval from Aug. 1885 to Dec. 1903. He finds that all the destructive earthquakes (16 in niunber) which visited Japan during this interval occurred exactly or very nearly at those epochs when the latitude of Tokyo was at a maximum or minimimi. The strong but non- destructive earthquakes (41 in number) show a similar ten- dency, though in a less marked degree. Of the slight or sensible earthquakes (303 in number), 180 occurred during 42 months in which the latitude was close to its maximum or minimum value, and 123 during the 28 months in which the latitude was increasing or decreasing. The corresponding rates per month were practically eqiial, namely, 4-3 and 4-4 respectively. Of moderately strong earthquakes, 78 occurred during 24 months of maximum or minimum latitude, and 54 during 17 months of decreasing or increasing latitude. The corresponding rates per month were again nearly equal, namely, 3-3 and 3-2. Thus, Omori concludes that destructive earthquakes in Japan have a marked tendency to occur in the epochs of maximum or minimum latitude at Tokyo, but that earthquakes of a moderate or slight degree of intensity are unaffected by the changes of latitude. The exact relation between the- changes of latitude and the earthquake-frequency (supposing further investigations should 198 FREQUENCY AND PERIODICITY [ch. xi prove its reality) is at present unknown. The crust-displacements in great earthquakes are insufficient to produce the observed deflections of the Pole, and it is inconceivable that deflections so minute should be the cause of the earthquakes. On the whole, it seems probable that both are rather due to the action of some common cause*. * J. Milne, Rep. Brit. Ass., 1900, pp. 107-108; 1903, pp. 78-80; 1906, pp. 97-99; A. Cancani, Boll. Soc. Sis. Itah, vol. 8, 1903, pp. 286-290; F. Omori, Publ. Eq. Inv. Com., No. 18, 1904, pp. 13-21; C. G. Knott, Rep. Brit. Ass., 1907, pp. 91-92. CHAPTER XII ACCESSORY SHOCKS 206. No great earthquakes, and few earthquakes of even moderate strength, oeeur alone. They are usually, but not always, preceded by sVight fore-sJwcks ; they are invariably followed by a number, generally a large number, of after-shocks. Both fore- shoeks and after-shocks are confined to the region containing or immediately surrounding the epicentre of the principal earth- quake. Besides these, there are others, known as sympathetic shocks, which occur in neighbouring districts but which are pre- cipitated by the occurrence of the principal earthquake and by the changed stresses which it suddenly introduces in the sur- rounding crust*. Fore-Shocks 207. Number and Intensity of Fore-Shocks. Fore-shocks, when they occur, are usually few in number and of slight in- tensity. Occasionally, they attain a semi -destructive strength. The great earthquake of central Japan, which occurred on July 9, ISo^, was preceded on the 7th inst. by two shocks at 1 p.m. and 2 p.m., strong enough to crack some plastered walls within the epicentral area of the ))rincipal earthquake. They were followed by incessant earth-soimds and by at least 27 minor shocks the same day. On July 8, the earth-sounds con- tinued, and there were a few small shocks at about 8 p.m. Six hours later the great earthquake occurred. As a rule, the fore-shocks are more insignificant. The Charleston earthtpiake of Aug. .31, 1886, was preceded on Aug. 27, at about 8 a.m., by a slight earthquake felt at Summerville, a village 22 miles to the north-west and not far from the principal epi- * On the accessory shocks of earthquakes, tlie followiiif^ [)ai)ers may be consulted: 1. Davison, C. On the distribution in space of the accessory shocks of tiie great .Japanese earthcjuakc of 1H!)1. (^luirt. Jouni. (ieol. Soc, vol. 53, 18«7, pp. 1-l.j. 2. Omori, F. On the after-shocks of earthquakes. Jourii. Coll. Set., Imp. Univ. Tokyo, vol. 7, 1894, pp. 111-200. 200 ACCESSORY SHOCKS [ch. centre.- On Aug. 28, at 4.45 a.m., another and stronger earth- quake oceurred at the same plaee and was felt as far as Charleston. During this and the following day, several other slight shocks -were noticed at Summerville, but -were succeeded by an interval of repose which lasted imtil 9.51 p.m. on Aug. 31, when the great earthquake occurred. Again, the first undoubted fore- shock of the Riviera earthquake of 1887, occurred on Feb. 22 at 8.30 p.m., about 10 hours before the principal earthquake, and this was followed by five others, one of which was per- ceptible all over the Riviera, and in Piedmont and Corsica. The principal earthquake occurred on Feb. 23, at 6.20 a.m. There appears to be no relation whatever between the occur- rence of fore-shocks and the strength of the principal earthquake. British earthquakes, which are never of great strength, are some- times preceded by perceptible shocks. Two fore-shocks of the Hereford earthquake of 1896, for instance, disturbed areas of 6300 and 6400 square miles, or about one-fifteenth the area shaken by the principal earthquake. On the other hand, the Messina earthquake of 1908 and the Californian earthquake of 1906, with its great fracture 270 miles in length, took place suddenly, without warning of any known kind. The Assam earthquake of June 12, 1897, one of the first rank of earthquakes with an epicentral area of 6000 or 7000 square miles, was pre- ceded, if at all. by tremors and earth-sounds so faint that, but for the subsequent earthquake, they would have passed un- recorded. It is possibly from the absence of seismographs or from want of careful observations, that fore-shocks seem to be rare or so often escape notice. Thus, it is said that the great Mino-Owari earthquake of 1891 was heralded by a strong shock 58 hours earlier, though earth-sounds were heard within the epicentral area during the intervening days. A more detailed examination of the previous earthquakes of the district shows, however, that a preparation for the principal earthquake had been going on for several years. 208. Distribution in Time of the Mino-Owari Fore-Shocks. The Mino-Owari earthquake occurred on Oct. 28, 1891, at 6.37 a.m. The extraordinary fault-scarp formed on that occa- sion is described in sect. 86. Its coiu'se is represented by the XII] ACCESSORY SHOCKS 201 continuous line in the accompanying map (Fig. 82), The dotted lines indicate the boundary of the meizoseismal area. The area shown in the map is bounded by the parallels of 34° 40' and 36° 20' X. lat., and by the meridians of 2° 10' and 3° 50' W. long, of Tokyo, so that, if the area be di^•ided by N.-S. and E.-W. lines one-sixth of a degree apart, there are ten rectangles ad- joining each boundary of the map and 100 rectangles within its ' *B \ \c" ; L • ; • : H • L • N \ ^ / ■■> '■■••■. J < J K Yk "\ ^~\'^ vi ^' 9 K Fig. 82. Course of tlie faidt-scarp, etc., of the Mino-Owari earthquake of 1891. area. The points .1-1* denote the centres of the rectangles in Avhich the majority of the fore-shock and after-shock epicentres lie. The points C and D, and possibly also E and F, seem to be connected with the main fault; A and B with the northern end of its fault-scarp, G and // with the southern end, and L and A'^ with a probable continiuition of the fault to the south- 202 ACCESSORY SHOCKS [ch. east. The central points K, M and P are probably connected with a deep-seated fault which follows the main branch of the meizoseismal area to the south, but of which no trace as a scarp is visible at the surface. The number of epicentres within the whole area of the map for the different years from 1885 to 1891 (to Oct. 27) varies from 16 to 52, and the number within the rectangles A-P from 4 to 32. Taking account of the fact that there are 13 rectangles along the course of the fault and 87 rectangles elsewhere, it follows that in 1885 there were 5-4 times as many centres near the faults as in an equal area elsewhere, in 1886 3-9 times, and in 1887 2-2 times. It is possible that this decline in relative frequency marks the decadence from the last strong earthquake in the Mino-Owari district, which occurred in 1859. After 1887, however, there is a rapid increase in the ratio. In 1888, it rises to 5-5, in 1889 to 7-0, and during the interval from Jan. 1, 1890, to Oct. 27, 1891, it rises still further to 10-5. The first symptom of the coming of the great earthquake is therefore an increase in the frequency of earthquakes along the lines of fault relatively to the frequency elsewhere*. 209. Distribution in Space of the Mino-Owari Fore-Shocks. Still more significant is the distribution of the fore-shocks of the Mino-Owari earthquake. During the five years 1885-1889, the deep-seated fault was almost entirely inactive, and, along the main fault, the epicentres Avere chiefly confined to the central region. During the 22 months before the earthquake, however, a remarkable change occurred in the distribution of the epi- centres. This is represented in Fig. 83, The central region of the main fault is still the principal seat of activity, but the curves now follow the courses of both faults. Earthquakes indeed occurred along the whole fault-system, especially along the line of the deep-seated fault and the continuation of the main fault towards the south-east. Moreover, they occurred with some approach to uniformity along the whole fault-region ; the marked concentration of activity which characterises the after-shocks (sect. 215) is hardly perceptible among the fore-shocks f. 210. Prevision of Earthquakes by means of Fore-Shocks. With' our present knowledge of the preparation for the Mino-Owari * Davison, pp. 9-12. f Davison, pp. 11-12, XIl] ACCESSORY SHOCKS 203 earthquake, the phenomena by which the oecurrence of the earthquake might have been foreseen are (i) the increase in frequency of fore-shocks from 1887 until October 1891 in the district surrounding the fault-system as compared with equal areas elsewhere; but especially (ii) the outlining of the fault- system by the curves of equal frequency of fore-shocks during the two years immediately preceding the earthquake. Fig. 83. Distribution of fore-shocks of the Minu-Owari earth(iuake of 1891. With a sufficient number of observing stations, the location of the epicentre of an earthquake is an easy matter. It is much simpler than the measurement of the displacement of pillars btiilt into the ground, as suggested by Reid. Moreover, it is possible that a great earthquake may, as in the Kangra earth- quake of 1905, occur without very perceptible distortion of the surface-crust*. * C. Davison, liiilr. lur (koph., vol. 12, 1012. pp. 9-l.j. 204 ACCESSORY SHOCKS [ch. After-Shocks 211. Number of After-Shocks. In striking contrast with the paucity of fore-shocks is the extraordinary abundance of the after-shocks. In some cases, it is no exaggeration to say that for days the epicentral area is never actually at rest. During the night following each of the earthquakes in north-east Greece on Apr. 20 and 27, 1894, the ground within the two innermost isoseismal lines was in a state of almost incessant disturbance. After the great Assam earthquake of June 12, 1897, the ground at several places in the epicentral area remained for days in a state of tremor, interrupted frequently by sensible shocks, and occasionally by severe earthquakes. At Bordwar, the surface of a glass of water was observed for a week afterwards to be continually trembling; and at Tura a lamp was kept swinging for three or four days. Without instrumental aid, all registers of after-shocks are inevitably incomplete. Indeed, when the trembling appears to be continual, separate after-shocks cannot be distinguished. It is only in rare cases when there are uninjured seismographs in the neighbourhood of the epicentre that we can form any idea of the actual frequency of the after-shocks. After the Mino- Owari earthquake of Oct. 28, 1891, the seismographs at Gifu and Nagoj^a, though overthrown, were re-erected within a few hours, the former recording 318, and the latter 185, shocks on the day after the earthquake. During the first 30 days, 1746 after-shocks were registered at Gifu, the total number by the end of 1893 amounting to 3365. After the Kumamoto earth- quake of July 28, 1889, 340 shocks were recorded at Kumamoto during the first 30 days and 833 during the first two years. The Kagoshima earthquake of Sep. 7, 1893, was followed by 278 after-shocks at Chiran within the first 30 days; the Hokkaido earthquake of Mar. 22, 1894, by 431 after-shocks during the first 30 days at Nemuro (74 miles distant from the epicentre) and 715 during the first year. Numbers approaching or exceeding those given above are the results of personal observations in several earthquakes. For instance, the Tenpo (Japan) earthquake of Aug. 19, 1830, was succeeded by 681 after-shocks at Kyoto in the first six months; the Zenkoji (Japan) earthqviake of May 8, 1847, by 930 shocks XII] ACCESSORY SHOCKS 205 at Matsushiro in the first 31 days ; and the great Japanese earth- quake of Nov. 4, 1854, by not less than 919 shocks before the end of the following year. In the first two of these earthquakes, the actual numbers of shocks probably rivalled that of the Gifu series of Mino-Owari after-shocks. After the Messina earthquake of Dec. 28, 1908, 949 after-shocks were counted at Messina by the end of 1909. Owing to the number and wide dispersion of independent centres of activity, the after-shocks of the great Assam earthquake of June 12, 1897, probably surpass in number those of any other known earthquake. In little more than three days, 561 shocks were felt in X. Gauhati. During one year, from Oct. 1, 1897, to Sep. 30, 1898, 1050 after-shocks were recorded at Maophlang and 841 at Mairang*. 212. Intensity of After-Shocks. The great majority of the after-shocks of an earthquake are of very slight intensity, the strongest as a rule being far inferior to the principal shock. Of the 3365 after-shocks of the Mino-Owari earthquake of 1891 recorded within a little more than two years at Gifu, 10 were violent, 97 strong, 1808 weak, 1041 feeble, and 409 were earth- sounds. The Riviera earthquake of Feb. 23, 1887, occurred at 6.20 a.m., and was followed at 6.29 a.m. and 8.51 a.m. by two shocks which added to the destruction and loss of life caused by the principal earthquake. After the great Assam earthquake of 1897, eight shocks were felt at Calcutta, 250 miles from the e})icentral area, one of them beyond Allahabad to a distance of more than 550 miles. Of the Jamaica earthquake of 1907, no fewer than 148 after-shocks were strong enough to be recorded by Milne seismogra})hs in Great Britain. The interest of after-shocks, however, lies not in their strength but in their weakness. For the most part, they are of local origin. The shocks felt at one jilace arc not usually the same as those felt at another only a few miles away. Thus, Maojjhlang and Mairang, two places in the epicentral area of the Assam earthquake (Fig. 47), are only 11 miles apart. During 17 days (Sep. 12-28, or three months after the earthquake), 92 after- * F. Omori, pj). 111-200; Publ. Kii- Inv. Com., No. 7, 1!)0'2, pp. ;{;j-51 ; liull. Eq. Inv. Com., vol. 2, 1908, |)p. 1H5-19.5; Boll. Soc. Sis. Hal., vol. 2, 18»fi, pp. 152-1.55; Oldham, pp. 124-128, and Mem. Geol. Surv. India, vol. .•{(». 1000, pp. 1-102. 206 ACCESSORY SHOCKS [ch. shocks were felt at Maophlang, 37 being described as smart, 45 slight and 10 feeble. In the same interval, 83 shocks were felt at Mairang, 6 being smart, 10 slight and 67 feeble. It is difficult to obtain correct time in Assam, but, regarding shocks as identical if their recorded times differ by not more than 15 minutes, there were in the interval referred to 19 shocks common to both places, leaving 73 as peculiar to Maophlang and 64 to Mairang. Moreover, of the 19 common shocks, only one was considered as smart at both places, 12 were smart at one and slight or feeble at the other, while 6 were slight at one and feeble at the other and may have been independent shocks. Again, the after-shocks of the Mino-Owari earthquake of 1891 were recorded by seismographs at six observatories, Gifu being 17 miles from the central portion of the Neo Valley where most of the shocks originated, Nagoya 37 miles, Tsu and Kyoto 61, Osaka 88, and Tokj^o 166, miles. The number of after-shocks recorded from Oct. 28, 1891 to Dec. 31, 1893, was 3365 at Gifu, 1298 at Nagoya, 314 at Tsu, 125 at Kyoto, 70 at Osaka and 30 at Tokyo. Omori has represented the local character of the after-shocks by means of curves drawn through all places at which the same number of shocks were felt. From the map for November 1891, he finds that 200 after-shocks were felt to a mean distance of 25 miles, 100 to a distance of 48 miles, and 10 to a distance of 112 miles*. 213. Decline in After-Shock Frequency. Niunerous as after- shocks are for a few days after a great earthquake, their decline in frequency is at first very rapid. The daily numbers recorded at Gifu during the first seven days after the Mino-Owari earth- quake of 1891 were 318, 173, 126, 99, 92, 81 and 78. A month later, the average daily number dwindled to 18. The decline in frequency of the after-shocks at Gifu is represented by the con- tinuous line in Fig. 84, in which the numbers of after-shocks in successive months from Nov. 1891 to Dec. 1893 are represented by the distances of the small crosses from the horizontal line. Thus, in Nov. 1891, there were 1087 after-shocks, in the next * F. Omori, Journ. Coll. Sci., Imp. Univ. Tokyo, vol. 7, 1894, pp. 112- 113; Oldham, p. 125; J. Milne, Rep. Brit. Ass., 1908, pp. 64-66; 1909, pp. 51-55; F. Omori, Publ. Eq. Inv. Com., No. 7, 1902, pp. 29-31. XIl] ACCESSORY SHOCKS 207 month 416. In Dec. 1892, the number had decreased to 39, and in Dec. 1893 to 16. The decHne in frequency, as will be seen from the curve in Fig. 84, is far from uniform. The succession of weak and feeble shocks was broken from time to time by violent earthquakes, each of which, as on Jan. 3 and Sep. 7, 1892, was followed by its own train of after-shocks and so gave rise to temporary fluctuations in the total number. The dotted curve in Fiff. 84 1892 1893 Fig. 84. Decline in frequency of after-sliocks of the Mino-Ovvari earthquake of 1891. is drawn so as to smooth away these irregularities. In all pro- bability, it represents the true law of decline in frequency, so far as the original earthquake is concerned. This dotted curve differs but little from a rectangular hyper- bola. Indeed, Omori finds that, in all the earthquakes which he has examined, its equation is of the form k where h and k are constants, and y is the number of earthquakes felt within a given interval at time x, measured from a fixed epoch. For the Mino-Owari after-shocks recorded at Gifu, Omori uses the half-dailv numbers of after-shocks during the 208 ACCESSORY SHOCKS [ch. five days Oct. 29-Xov. 2, 1891. Inserting for y the number of after-shocks during the 12 hours denoted by x, measured from the first half of Oct. 29, he thus obtains ten equations for finding h and /;, the values of which are determined by the method of least squares, the resulting equation being _ 440-7 ^ ~ x+ 2-31 ■ Though this equation is obtained from the nvmibers of after- shocks recorded during the first five days, it has nevertheless been used by Omori for determining the numbers of after- shocks felt after the lapse of six to eight years. Taking account of the usual annual number of earthquakes (18-3) recorded at Gifu, Omori estimates that the number of earthquakes in the two years 1898-1899 should be 160. The number actually re- corded was 163*. 214. Decline in After-Shock Intensity. The decline in the intensity of after-shocks, like their decline in frequency, is rapid at first and fluctuating. Of the numerous after-shocks of the Assam earthquake of June 12, 1897, eight were strong enough to be felt in Calcutta, and all of them occurred within the first four months. The after-shocks of the Kumamoto earthquake of July 28, 1889, included one violent earthquake on Aug. 3, and 76 strong shocks, 29 of which occurred before the end of July, and the last on May 28, 1890. The Hokkaido earthquake of Mar. 22, 1894, Avas followed at Nemuro by 11 strong shocks, of which eight occurred before the end of March, two in April, and one on July 14. Of the 10 violent after-shocks of the Mino- Owari earthquake of Oct. 28, 1891, nine occurred within the first four months, and the last on Sep. 7, 1892. All of the 97 strong shocks occurred within the first 13 months, and all the weak ones but four within the first 20 months. Towards the end of 1893, besides the four weak shocks, only feeble shocks and earth-sounds were observed. * F. Omori, p. 118, and Piibl. Eq. Inv. Com., No. 7, 1902, pp. 27-29. It should be mentioned that, as regards the Riviera earthquake of Feb. 23, 1887, and the Messina earthquake of Dec. 28, 1908, Cavasino and Aga- mennone find that the decline in frequency of the after-shocks does not follow Omori's law {Boll. Soc. Sis. Hal., vol. 15, 1911, pp. 129-143; RivUia di Aslronomia, etc., Nov. 1912). The explanation probably is that both of these earthquakes were twin earthquakes (see sect. 216). XIl] ACCESSORY SHOCKS 209 100 80 60- 40 1 r ^ \ r Y v^ / " V / / / ■ 20 1892 1893 Fio. 85. Percentage of feeble after-shocks of the Mino- Owari earthquake of 1891 at Gifu. Fig. 86. Distribution of after-shocks of the Mino-Owari earthquake of 1S!)1 (Nov .-Dec. 1891). 14 210 ACCESSORY SHOCKS [CH. The fluctuating, but on the whole increasing, proportion of feeble shocks registered at Gifu is represented by the curve in Fig. 85, which shows the varying percentage in successive months of the feeble shocks with regard to the total number of shocks and earth-sounds. Fig. 87. Distribution of after-shocks of the Mino-Owari earthquake of 1891 (July-Aug. 1892). 215. Focal Migration of After-Shocks. With the lapse of time, after-shocks not only decline in frequency, but also vary in the position of their origin. As in the case of fore-shocks, our knowledge is chiefly based on the after-shocks of the Mino- Owari earthquake of Oct. 28. 1891. The maps in Figs. 86 and 87 are constructed in the same manner as the map showing the distribution of the fore-shocks (Fig. 83). The first map (Fig. 86) shows the distribution of the after- XII] ACCESSORY SHOCKS 211 shock epicentres in November and December 1891, the two months immediately following the great earthquake. The second (Fig. 87) shows the distribution eight months later, in July and August 1892. A comparison of the maps shows at once the decline in frequency; in Fig. 86 there is a curve corresponding to 1000 epicentres, in Fig. 87 the curve of highest order is one corresponding to 60 epicentres. In November and December 1891, after-shocks occurred at the northern end of the fault- scarp, but chiefly in the central and southern regions*. The epicentres cluster along the continuation of the fault-scarp to- wards the south. The greatest activity, hoAvever, is concentrated along the secondary fault, of the existence of which evidence has been already given (sect. 209). Fig. 87 shows that, eight months later, seismic activity had abandoned the terminal regions of the fault. A few epicentres remain near the southern end of the fault-scarp; the larger number are grouped around the more central portion of the scarp, perhaps also along the secondary fault, but in a more northerly portion of it than in the months following the principal earthquake. The distribution of after-shock epicentres thus points to the existence of a nearly central region of great activity with regions of minor activity near or surrounding the ends of the faults. The activity of these terminal districts was not only less marked, it was also of shorter duration, than that of the central region. At the northern end of the main fault and at the south-eastern end of its continuation, all activity had practically ceased before April 1892. In the region surrounding the southern end of the fault-scarp it lasted until about the close of the same year. A similar withdrawal took place from the southern end of the secondary fault, only two epicentres lying in that district after March 1892. Thus, the distribution of after-shock epicentres of the Mino-Owari earthquake is marked by decrease in the area of activity, and by its gradual but oscillating withdrawal to a more or less central district t. 216. After-Shocks and the Nature of the Principal Earthquake. * Tlie break between the northern and central groups of eurves (Fij^. 80) may be apparent rather tlian real, for ti»e country in tiiis part is moun- tainous. + Davison, pp. 5-14. Other diagrams showing the distribution of after- shock epicentres are given on pp. 5-7 of this paper. 14r— 2 212 ACCESSORY SHOCKS [ch. The relations between the frequency of after-shocks and the nature of the principal earthquake cannot yet be considered in detail. Two points, however, are worthy of notice. (i) There seems to be a marked difference in the nimiber of after-shocks which attend twin and other earthquakes. In Great Britain, the earthquakes are invariably either simple or twin. During the 21 years 1889-1909, the three simple earthquakes of Inverness in 1890 and 1901 and Carnarvon in 1903 were followed by at least 33 after-shocks, w^hile 7 twin earthquakes (those of Pembroke in 1892 and 1893, Hereford in 1896, Derby in 1903 and 1904, Doncaster in 1905 and Swansea in 1906) were followed by 18, the ratio per earthquake being 11 to 2-57 or more than 4 to 1. The same relation seems to govern the stronger twin earthquakes of other lands. The Charleston earthquake of 1886 was attended by comparatively few after-shocks; the Riviera earthquake of 1887 by 673 after-shocks from Feb. 23 to Dec. 31, 1887. The Messina earthquake of 1908, with its 949 after-shocks in little more than a year, seems to form an exception to the relation indicated. In this case, however, the focus was evi- dently close to the siu'face and changes of elevation w^ere mea- sured along the coasts of the Straits of Messina (sect. 89). (ii) There is an evident relation between the number of after- shocks of an earthquake and the height of the fault-scarp. In the Mino-Owari earthquake of 1891 and the Assam earthquake of 1897, the fault-scarps were in places of considerable altitude (sect. 81) and the numbers of after-shocks were unusually large. The great Concepcion earthquake of 1835 was probably caused by a displacement along a submarine fault nnming in a direction parallel to the neighbouring coast-moimtains. That it was ac- companied by the formation of a fault-scarp is clear from the occurrence of the sea-wave which afterwards swept over the adjoining shores. The coast was also raised by several feet, at one point by not less than 10 feet. The earthquake was followed by hundreds of after-shocks, some of considerable violence, pro- ceeding apparently from the same origin. At the same time, the coast subsided, for, after an interval of some Aveeks, it stood at a lower level than it did immediately after the principal earthquake. A large number of after-shocks of the Assam earthquake of XII] ACCESSORY SHOCKS 213 1897 were recorded at Maophlang, where an interesting observa- tion was made. A straight piece of wood was nailed to a stout post so that its upper edge pointed exactly to the crest of a ridge about a mile and a half to the west. Six months later, this edge pointed some way down the slope of the ridge, being apparently tilted through an angle of one degree. The change might be due to a displacement of the post, of which, however, there was no evidence. It probably implies that crustal dis- placements continued long after the great earthquake, and that they were, in part at any rate, due to the movements which caused the stronger after-shocks. Lastly, the number of after-shocks of the Californian earth- quake of 190G was extraordinarily small, if wc consider the vast length of the fractiu-e and the volume of the displaced crust (sects. 80, 84). During the first 14 months, the total number of recorded shocks in no place exceeded 153. Xor. within the central area, was there the incessant quivering which is so common a feature in the early days after a great earthquake. Indeed, as regards the mmiber of its after-shocks, the Californian earthquake was inferior to a disturbance so slight comparatively as the Comrie earthquake of 1839. In the Californian earth- quake, however, the displacement, great as it was. was mainly horizontal. A perceptible fault-scarp was confined to the northern half of the epicentral area, and it was in this district that after- shocks were more frequent than elsewhere. The paucity of Californian after-shocks was no doubt due to the small range of the vertical displacement*. Sympathetic Shocks 217. Sympathetic Shocks of the Mino-Owari Earthquake. The stresses to which the crust is subjected before and after a great earthquake are not confined to the region of the fault aloiic. In the whole surrounding country, they must be different after a great earthquake from what they were before. They may be increased or decreased by the displacement which produced the earthquake, and the result may be either an increase or decrease in the seismic activity of the neighbouring regions. So far as * C. Davison, Geol. Mag., 1910, pp. 417-418; A. Cavasino, Boll. Soc. Sis. Hal., vol. \->, 1911, pp. 142-14.3; C. Darwin, Trans. Geol. Soc, vol. 5, 184(), pp. 018-G19; Oldliam, pp. 157-1.)8; Lawson, vol. 1, pp. 41(>-4:{;J. 214 ACCESSORY SHOCKS [ch. xii regards the districts which surround that in which the Mino- Owari earthquake originated, the result was a marked increase in activity. Abovit 45 miles to the east, and 55 miles to the west, of the great fault-scarp are two other districts in which earthquakes are somewhat frequent. In the eastern district 29 earthquakes, and in the western district 20 earthquakes, originated between Jan. 1, 1885, and Oct. 27, 1891. After the earthquake, from Oct. 28, 1891, to the end of 1892, the numbers w^hich originated in the same districts were 30 and 36, in Nov. 1891 alone 7 and 8. Thus, for every earthquake in the eastern district in the period before 1891, 6 were felt in the interval afterwards and 10 in the month of November 1891 alone; for every earthquake in the western district before Oct. 1891, 10 were felt in the interval afterwards, and 16 in November 1891 alone. The marked increase of seismic activity in these two districts need not, however, be a consequence of the great displacement. The shocks in both central and adjacent districts, it is possible, might result from a general increase of stress over a wide extent of country, and the augmented frequency in the lateral districts could not with justice be regarded as an effect of the former, for they might both be effects of the same widely prevailing- cause. But that the connexion is one of real dependence is probable for two reasons, (i) Crustal distortions of the kind and magnitude of those which took place in the Neo Valley could not be effected without a very considerable change of stress in all the surroimding country, (ii) An increase of stress cannot determine the occurrence of an earthquake unless it be svifficient to overcome the resistance to effective displacement. Now, it is luilikely that the gradual increase of stress should be so nearly proportioned everywhere to the prevailing condi- tions of resistance as to give rise to a marked and practically simultaneous change in seismic activity over a large area; whereas the sudden occurrence of a strong earthquake might alter the surrounding conditions with comparative rapidity, and induce a state of seismic excitement in the neighbourhood. The rapid and simultaneous increase in earthquake-frequency in the two subsidiary districts, distant though they be from one another by 100 miles, seems strongly in favour of this interpretation*. * C. Davison, Geol. Mag., 1897, pp. 23-27. CHAPTER XIII VOLCANIC EARTHQUAKES Relations between Tectonic and Volcanic Earthquakes 218. Earthquakes have been divided into two main classes (sect. 8) — tectonic and volcanic earthquakes. Tectonic earth- quakes are due as a rule to the displacements which effect the growth of faults. Volcanic earthquakes may be defined as those which are caused by the operations which result or tend to result in a volcanic eruption or are due to displacements, by whatever cause they may be produced, along fractures of the volcanic mass, whether the volcano itself be active, dormant or extinct. Thus, volcanic earthquakes are of two kinds : (i) those which are purely volcanic in their origin, and (ii) those which are of tectonic origin in so far as they are due to the growth of faults, but of volcanic origin in that the slips are precipitated by present or past volcanic operations. 219. General Independence of Tectonic Earthquakes and Vol- canic Eruptions. A small scale seismic map of the world, such as that of Mallet, conveys the impression that active volcanoes are situated as a rule in regions in which earthquakes are numerous and strong. More detailed maps lead to a different conclusion. For instance, Milne's seismic map of Japan, in which epicentres alone arc indicated (sect. 173), shows that "the central portions of Japan, which are the mountainous districts where active volcanoes are numerous, are singularly free from earthquakes." This feature is brought out still more clearly by the map of Japan in Fig. 70. The volcanoes, most of which are active, are represented by black dots, and these, it will be seen, arc almost entirely absent from the more darkly shaded areas. Excluding those in the smaller outlying ishnids, there are 88 volcanoes in Japan. Some of these lie exactly on the border of two of the rectangles into which Milne divides the country (sect. 172), and there are thus 96 rectangles in which volcanoes arc entirely or partly situated, while there are 1476 rectangles 216 VOLCANIC EARTHQUAKES [CH. .S^ o on land without volcanoes. Now, of the Japanese earthquakes of the years 1885-1892, the average number of epicentres in each non-volcanic rectangle is 3-57, and in each volcanic rect- angle 0-73; so that the number of epicentres in a non-volcanic rectangle is very nearly five times as great as in a volcanic rectangle. Somewhat similar relations govern the occiu'rence of earth- quakes and volcanic eruptions with regard to time. If long- intervals of time, one or two centuries, be considered, it appears, as Mercalli has shown, that the intervals, 1632-1737, 1750-1849, in which earthquakes were numerous and strong were also those in which eruptions of Vesuvius were frequent; while the intervals, 1303- 1499, 1503-1631, in which earthquakes were of less consequence were also those during which Vesuvius was seldom in eruption. If short intervals of time be con- sidered, this general coin- cidence, as will be seen later, disappears, and the earth- qviakes as a rule precede or follow the enqitions and only rarely accompany them*. 220. A remarkable exception to the last statement, in which volcanic eruptions are occasionally accompanied by true tectonic earthquakes, remains to be noticed. For instance, on Jan. 12, 1914, a few hours after the great eruption of the Sakura-jima (south Japan) began, there was an earthquake strong enough to damage houses within a few miles of the volcano and to be Scale of Mjlf 10 20 30 40 50 Fig. 88. Volcanic chain of south Japan. * R. Mallet, Rep. Brit. Ass., 1858, plate 12; J. Milne, Sets. Journ., vol. 4, 1895, p. XV ; C. Davison, Geogr. Journ., vol. 10, 1897, p. 534; G. Mercalli, Vulcani e Fenomeni Vulcanici in Italia, 1883, pp. 357-359; J. Milne and H. H. Turner, Rep. Brit. Ass., 1913, pp. 65-67. xiii] VOLCANIC EARTHQUAKES 217 recorded in European observatories. One month later, on Feb. 1 3, a similar earthquake took place during the cruj^tion of the Iwo- jima, a ^ olcano belonging to the same chain as the Sakura-jima. The course of this chain is shown in Fig. 88. On Nov. 18, 1913, the Kirishima-yama broke out in strong eruption; the Sakura- jima followed about two months later; and, one month after that, and still farther to the south, the Iwo-jima. When three volcanoes, situated as these are and all of infrequent activity, burst into eruption so nearly together, and, when two of the eruptions are accompanied by strong and deeply-seated earth- quakes, it is difficult not to regard both phenomena as different manifestations of a common cause, namely, the gradually gro^^^ng stresses along the whole volcanic chain. But there is no reason for supposing that the earthquakes result from the volcanic operations. They should therefore be considered as tectonic, and not as volcanic, earthquakes*. Eaktuqiakes of Active Volcaxoes 221. Etnean Earthquakes. Earthquakes occur on all sides of Etna. but. for some time, they have been specially frequent and violent on its south-eastern flank. Brief descriptions will now be given of a few typical earthquakes. A great eruption on the east-north-cast flank of Etna began on Sep. 10, 1911, and lasted only 13 days. Three weeks later, on Oct. 15, occurred the destructive earthquake of Fondo Macchia. which was preceded by at least ten fore-shocks and followed by five after-shocks, the whole series lasting from Sep. 30 to Nov. 9. The isoseismal lines of the principal earthquake, corresponding to intensities 10, 8, 6, 4, 2 of the Mercalli scale are shown in Fig. 89. The meizoseismal area (bounded by the isoseismal 8) is a slightly sinuous band. 4 miles long, about J of a mile wide, and 1{ sq. miles in area. Notwithstanding its great intensity within this band, the shock was not felt at places 6 miles to the west, and the area within the isoseismal 4 was not more than 70 sq. miles. The dotted line on the same map represents the boundary of * G. P. Scrope, Considerations on Volcaiwa (182.j), p. 155; F. Omori, Bull. Eq. Inv. Com., vol. 8, 1914, pp. 23-24; Nature, vol. 92, 1914, pp. 716- 717. 218 VOLCANIC EARTHQUAKES [CH. the meizoseismal area of the Fondo Macchia earthquake of July 19, 1865, also a narroAV band, 5 miles long, Ij miles wide, and containing about 5 sq. miles. As the shock was felt at no place more than 12 miles from the epicentre, its disturbed area must have been less than 113 sq. miles. It was followed by a number of after-shocks, the last of which occurred on Aug. 23 Scale or illles Fig. 89. Map of the Fondo Macchia earthquakes of 1865 and 1911. This earthquake also succeeded a great eruption of Etna, which began on Jan. 30, 1865, and lasted for nearly 12 weeks. The direction from Fondo Macchia of the central crater of Etna is represented by the arrow in Fig. 89. One of the most remarkable series of Etnean earthquakes occurred in the neighbourhood of Linera on May 8, 1914. The total number of sensible shocks was 55, 21 being fore-shocks from Apr. 28 to May 7, and 33 after-shocks from May 8 to XIIl] VOLCANIC EARTHQUAKES 219 June 4. The curves in Fig. 90 represent the boundaries of the areas within which houses were damaged by the more important shocks, A and B of the double earthquake of May 7 at 6.35 p.m., C of another earthquake on the same day at 10 p.m., the fracture within it being probably a continuation of that within the area B, D and E of the principal earthquake of May 8, and F that of the after-shock of May 26. Within the area bounded by the 2 4 Fi;,'. no. Map of tlic Lincra (Etna) earthquakes of May 8, 1914. cur\e D, which is about 4^ miles long and 1^ miles wi(k'. not only were houses completely razed to the ground, but the ground itself was crushed. Along the axis of this zone, there runs a slightly sinuous fracture, IVom Passopomo through Linera to the sea, its course being represented l)y the broken line on the map. In ahnost every part of it, there is a change of level, of an inch or two only in some phices, in others of 15 or 16 inches, and in one, near the sea-coast, of more than 3 feet, the 220 VOLCANIC EARTHQUAKES [CH. ground on the south-west side being left at the higher level. The direction of the central crater from Linera is represented by the arrow in the map*. Thus, the longer axes of the isoseismal lines of the earthquakes of 1865 and 191 i are directed toAvards the central crater, while those of the earthquake of 1911 are nearly at right angles to that direction. These earthquakes are types of two classes of ( > Paterno Scale or :Aile 10 Fig. 91. Seismic districts of Etna. Etnean earthquakes, the majority of which are probably con- nected with radial fractures of the volcano, and others with peripheral fractures'!". 222. Most Etnean earthquakes, as in the above examples, * During the interval covered by this series of earthquakes, there was a marked increase in the activity of Etna, though the shocks were not coincident with volcanic explosions. t A. Ricco, BoU. Soc. Sis. Ital., vol. 16, 1912, pp. 9-32; M. Baratta, Boll. Soc. Geogr. Ital., Oct. 1894; G. Platania, Pubbl. dcW 1st. di Geogr. Fis. e Vulcan, delta R. Univ. di Catania, No. 5, 1915. xiii] VOLCANIC EARTHQUAKES 221 disturb very small areas, and the villages in which houses are damaged are in all such cases close to the epicentres. Baratta has shown that Etneau earthquakes are especially frequent in 12 zones, which he names after neighbouring towns or villages. These are given in the accompanying map (Fig. 91), the dotted lines joining several places (such as Aderno, Bronte and Maletto) indicating that "their environs form a single zone. Though the zones on the south-eastern flank of Etna are at present those which are most frequently in action, earthquakes are by no means confined to them, and are indeed subject to frequent transferences from one zone to another. For instance, in one year only (1903), 8 of the 12 zones were in action*. In any one zone, the epicentres are not absolutely fixed, though, time after time, the same village in a zone {e.g. Xicolosi or P'ondo Macchia) may be damaged or destroyed f. 223. Japanese Volcanoes. Some interesting observations with a horizontal-pendulum seismograph have been made by Omori on the minute tremors which accompany volcanic eruptions in Japan. The Asama-3'ama (central Japan), which rises to a height of 8140 feet above the sea, was subject to a series of strong explo- .sions during the years 1908-1914. The observations were made in the summer months ^t a station, 6306 feet high, on the south- western flank. The seismograms arc divided by Omori into two classes, (i) those due to earthquakes which were not accompanied by any outburst of the volcano, and (ii) those due to earth- quakes which were invariably coincident with explosions. The former consisted only of minute quick vibrations; the latter began with slow movements on which, after a few seconds, quick vibrations were superposed. The earthquakes without explosions were distinctly the stronger — of the 1485 shocks re- corded from 1911 to 1916, 21 per cent, were sensible; while of the 8847 earthquakes with explosions, only 0-3 ]X'r cent, were * Nariuly, the zone of Lin (4 shocks); of Trecasta and June 1-1 <> (2;{ shocks); of S. Venerina, etc., on Mar. 11 and Nov. 20: anj of (;iarre-Hi|)osto on .Ian. :W. t M. Haratta. / Terremati (C Italia. litOl. p|). «2«-H:{:5: S. .Vrcidiacono, Itiill. iltir Accad. (iioeiiia ili Sci. Sal., in Valania, Kasc. 79, 1903. 222 VOLCANIC EARTHQUAKES [ch. sensible. Again, the two types of earthquakes alternate in fre- quency, the maxima of one type occurring at about the same time as the minima of the other*. 224. The last eruption of the Usu-san (north Japan) occurred in 1910, and consisted of explosions in a series of craterlets arranged along a peripheral fractiu-e of the volcano. Seismo- metric observations were made at two places, one (West Kohan) close to the east end of the line of craterlets, the other (Nishi- Monbets) 5 miles from the crater. At the former station, but not at the latter, series of well-defined minute quick vibrations, called micro-tremors by Omori, were recorded. The mean range of motion was always less than one-tenth of a millimetre, and the principal periods (-53, 1-08, 1-59 and 2-14 seconds) were practi- cally identical with those of earthquake-vibrations registered at Nishi-Monbets. Omori therefore concludes that micro- tremors are true earth- vibrations, but so w^eak that they cannot be recorded more than a few miles from the origin. Violent exjDlosions in the craterlets were usually accompanied, and sometimes preceded, by micro-tremors f. Earthquakes of Dormant and Extinct Volcanoes 225. Ischian Earthquakes. Ischia is a small island 6 miles from the west coast of Italy and about 20 miles from Naples. The central crater of M. Epomeo has long been extinct, but an eruption occurred from a lateral cone in the year 1302. The volcano may therefore be regarded as dormant. From this year until 1796, the island was practically free from earthquakes. A series of earthquakes then began, strong enough to damage houses in Casamicciola, followed by others in 1828, 1841, 1867, 1881 and 1883, by the last of which Casamicciola was ruined. In the map (Fig. 92), the dotted lines represent the boundaries of the existing portions of Epomeo, and the continuous lines the boundaries of the area within which buildings were seriously damaged by the earthquakes of 1796, 1828, 1881 and 1883. The broken line shows the position of the radial fracture with which the earthquakes were connected, and it was along this line, in the neighbourhood of Casamicciola, that the chief damage in 1881 and 1883 was concentrated. * F. Omori, Bull. Eq. Inv. Com., vol. 7, 1917, pp. ii-iii. t Ibid., vol. 5, 1911, pp. 31-38. XIIlJ VOLCANIC EARTHQUAKES 223 In 1881, as Johnston-Lavis has shown, the area of complete destruction included only half a sq. mile, the area of serious damage about 2 sq. miles, and that of partial damage about 5 sq. miles, while the shock was just felt on the Italian coast, which lies about 10 miles from Casamicciola. In the much stronger earthquake of 1883, the areas of complete destruction, serious damage and partial damage were respectively 3, 11 and 30 sq. miles, while the shock was felt b}^ a few persons in Naples. In both earthquakes, the depth of the focus, as estimated from the inclination of fissures in buildings, was found to be about , 1011, pp. 8-17; vol. 8, HM4, pp. 9-14, 22-27. D.M.S. 16 226 VOLCANIC EARTHQUAKES [ch. (v) Accessory Shocks limited in Time and Space. Some, but not all, volcanic earthquakes are preceded and followed by accessory shocks. In this, they resemble tectonic earthquakes. But the after-shocks of volcanic earthquakes are distinguished by the short period of their action. For instance, the after- shocks of the Albano earthquake of 1906 lasted for 3 days, in the Ischian earthquake of 1883 for 7 days, in the Fieri earth- quake of 1914 for about 3 weeks, in the Linera earthquake of 1914 for nearly 4 weeks, and in the Fondo Macchia earthquake of 1865 for 5 weeks, Fieri, Linera and Fondo Macchia being on the south-eastern flank of Etna. x\gain, the after-shocks of volcanic earthquakes are practically confined to the epicentral area. They point to little, if any, tendency towards an extension of the original focus. • (vi) Stability of Epicentre. In a volcanic system, such as that of Etna or the Alban Hills, earthquakes occur in many different zones and seismic activity is subject to frequent and sudden migrations from one zone to another. Nevertheless, in any given zone, there is often a certain fixity in the epicentres of successive earthquakes. In the Alban Hills, villages such as Frascati, Albano or Ariccia are shaken, while the surrounding- country is almost undisturbed. Since the beginning of the 14th century, every earthquake of any consequence in Ischia has originated in the same district close to Casamicciola. In the Etnean zones, time after time, the same places, such as Nicolosi, Fondo Macchia, etc., are damaged or destroyed. 228. The feature in which volcanic earthquakes differ most widely from tectonic earthquakes is the great intensity of the shock near the centre of a very small disturbed area. That this rapid decline in intensity is due to the shallowness of their foci is clear from sect. 139; and this inference is supported by two other lines of evidence. In the Ischian earthquakes of 1881 and 1883, the depth of the focus was estimated to be one-third of a mile (sect. 133). Again, the duration of the preliminary tremor of an earthquake increases with the distance from the focus, and the brevity or practical absence of any such tremor in volcanic earthquakes shows that the foci must be very close to the surface. 229. We may thus conclude: (i) that the foci of volcanic xiii] VOLCANIC EARTHQUAKES 227 earthquakes are very shallow; (ii) that the foci are usually small and not often more than 4 or 5 miles in length; (iii) that the accessory shocks are practically confined to the original focus; and (iv) that, while most volcanic earthquakes originate along radial fractures of the mountain, some — and those not the least important — originate along peripheral fractures. Origin of Volcaxic Earthquakes 230. The earthquakes which occur near or below active and dormant volcanoes are naturally attributed to the pro- cseses which tend to result in a volcanic eruption, the causes appealed to being: (i) the formation of new fractures, or the re-opening or extension of old fractures, in the mountain-mass; (ii) explosions, due to any cause, within the volcano; (iii) the sudden injection of lava into fractures or cavities in the mass of the volcano; and (iv) the relative displacement of the rock- masses adjoining a fracture of the volcano. * In active volcanoes, there can be no doubt that the formation or extension of fractures and (especially) explosions give rise to a large number of very slight earthquakes. But, as we know from the evidence of the Japanese seismograms, the earthquakes accompanying explosions are weak and the great majority im- perceptible to the unaided senses. Nor is there any reason for supposing that the process of fracturing would cause shocks of much strength, and, in any case, the frequent repetition of shocks in the same region would be difficult to explain on this theory. The sudden injection of lava into fractures or cavities of the mountain would be a more efficient cause of strong earth- quakes, especially beneath the flanks of a dormant volcano like M. Epomeo. The theory may be held to account satisfactorily for many of the phenomena of volcanic earthquakes; it is hardly applicable to the earthquakes of an extinct volcanic district. On the whole, it is probable that the more important volcanic earthquakes are due to the relative dis})laccmcnt of the rock- masses adjoining a fracture of the volcano. Wc know, from the evidence of the Linera earthquake of 1914, that such displace- ments do occur; it is clear that the resulting friction, depending as it does on the weight of the mass displaced, must be 15—2 228 VOLCANIC EARTHQUAKES [CH.- capable of producing such strong earthquakes as those which visit the flanks of Etna and Epomeo. 231. In some of the mining districts of Great Britain, there are occasionally earth-shakes which, though of much less strength than the Etnean earthquakes, resemble them closely in their nature, and j^robably also in their origin. In Fig. 94 are shown the isoseismal lines of an earth-shake which occurred at Pendleton (near Manchester) on Nov. 25, 1905. The shock was of intensity 7 (Rossi-Forel scale) near the centre of a disturbed Fig. 94. Map of the Pendleton earth-shake of Nov. 25, 1905. area containing 144 sq. miles. As the average disturbed area of British earthquakes of the same intensity is 24,500 sq. miles, it is evident that the earth-shake originated at a very slight depth. The mean direction of the longer axes of the isoseismal lines is N. 37° W., that of the Irwell Valley fault near the epi- centre (represented by the broken line in Fig. 94) is N. 34° W. It therefore seems probable that the earth-shake was due to a slip along this fault, caused either by the pumping of water from the mine or by the removal of the coal up to the face of the fault. If so, the earth-shake was of natural origin in so far xiiij VOLCANIC EARTHQUAKES 229 as it was due to the growth of the fault, but of artificial origin in that the slip was precipitated by mining operations*. 232. An active volcano is traversed by many fractures, the majority of which are radial, but a few are peripheral. If one or both of the masses adjoining a fractiu'e were to be deprived of support, the resulting displacement would give rise to an earthquake, not to be distinguished, except by its scale, from a true tectonic earthquake. The support might be withdrawn cither by underground movements of the magma f, or, less fre- quently, by the cooling of a mass of lava or heated rock below, by the former in active volcanoes, and by the latter especially in dormant and extinct volcanoes. This theory, it will be seen, accounts for all the known phenomena of volcanic earthquakes — for their close connexion in space and time with many eruptions, the shallowness and small size of the foci, the frequent repetition in the same region, the intensity and brevity of the shocks, and the occurrence of series of fore-shocks and after-shocks of brief duration and limited zone of displacement. * C. Davison, Geol. Mug., 1905, pp. 219-223; 1906, pp. 171-176. t G. Platania, Pnbbl. deir 1st. cli Geogr. Fis. e Vulcan, della R. Univ.di Catania, No. 5, 1915, pp. 1-2, 41. CHAPTER XIV ORIGIN OF TECTONIC EARTHQUAKES Earthquakes and the Growth of Faults 233. An earthquake has been defined as the result of any sudden displacement within the earth's crust (sect. 1). Of the displacements known or inferred, the following have been suggested as possible causes of earthquakes : (i) The fall of rock in iniderground channels; (ii) The operations which result or tend to result in volcanic eruptions; connected with which are (iii) Explosions of suddenly generated steam when water filtering through the outer crust reaches the highly-heated rock below ; (iv) The fracturing of the solid crust; and (v) The intermittent growth of faults, the usual cause being the friction generated by the sudden sliding of one rock-mass against the other, but complicated sometimes, when the dis- placement extends to the surface, b}^ the movement of the mass as a whole*. To the first of these causes we may perhaps attribute some slight local shocks. The earthquakes connected with volcanic eruptions have been considered in the preceding chapter. Often destructive within a limited area, they are seldom felt more than a few miles from the origin, and the greatest of all earth- quakes occur in regions which are far removed from present or * The following references may be given to memoirs in which the above theories are considered at greater length than is here possible: 1. Davison, C. (1). Twin-earthquakes. Quart. Jour n. Geol. Soc, vol. Gl, 1905, pp. 18-33. 2. (2). The Origin of Earthquakes (Cambridge Univ. Press), 1912. 3. Lebour, G. A. On the breccia-gashes of the Durham coast and some recent earth-shakes at Sunderland. Trans. N. of Eng. Inst, of Min. Eng., vol. 33, 1884, pp. 165-174. 4. See, T. J. J. The cause of earthquakes, mountain formation and kindred phenomena connected with the physics of the earth. Proc. Amer. Phil. Soc, vol. 45, 1906, pp. 274-414. CH. XIV] ORIGIN OF TECTONIC EARTHQUAKES 231 past volcanic action. To account for such earthquakes, the third cause has been invoked. We ha-ve no proof, however, of the occurrence of such explosions, nor does the theory provide a complete explanation of the distribution and phenomena of earthquakes. Ki. Portion f)f tlic nicizoseismal area of tlic Californian caithquako of Aj)!'. IS. lOOO. The principal reasons for comiectin^- eartli(|uakes with fault- slips are the foUowinfj: (i) with every step in the growth of a fault, it is evident that an eartlKpiake must occur: (ii) in some fjreat earthquakes, the fault-displaecinents are manifest; (iii) in all but the weakest earthquakes. Ilu- inner isoseismal lines are 232 ORIGIN OF TECTONIC EARTHQUAKES [ch. elongated in form, their longer axes being parallel to the fault- lines of the district : (iv) the nimiber of earthquakes in any region far exceeds the number of the faults ; (v) in a series of associated earthquakes, the epicentre migrates to and fro in the direction of the fault; and (vi) owing to variations in the volume and displacement of the rock-mass, fault-slips are capable of pro- ducing the weakest tremor as well as the most violent shock. 234. Fault-Displacements. In the somewhat rare cases in which fault-displacements occur (see Chapter V) the connexion between the origin of the earthquake and the groAvth of the fault is evident. In some cases, as in the IMino-Owari earthquake of 1891 and the Californian earthquake of 1906, it is further shown by the manner in which the inner isoseismal lines cling- to the course of the fault. In Fig. 95, for instance, the central continuous line represents a portion of the San Andreas fault (sect. 84) to the south of San Francisco. The darkly-shaded area is that which lies within the isoseismal 10 (Rossi-Forel scale), and the lightly-shaded area that which lies within the isoseismal 9. In no part of its course, which includes the entire land-area of the fault-displacement, does the width of the inner band exceed 2 miles; and, throughout its whole extent, the fault-line runs almost centrally between its boundaries. That the earthquake in such cases is to be attributed to the growth, rather than to the formation, of the fault, is evident from the fact that in some earthquakes — such as those of Baluchistan in 1892, Alaska in 1899, and Cahfornia in 1906 — the movement has taken place along pre-existing faults. 235. Elongated Forms of Isoseismal Lines. The forms of the isoseismal lines of some British earthquakes are shown in Figs. 28, 30, 50, 51 . In these cases, the inner curves are elongated, while the outer curves are nearly circular in form. In the Hereford earthquake of 1896, the dimensions of the innermost isoseismal line are 40 and 23 miles ; in the Inverness earthquake of 1901 (Fig. 50) 12 and 7 miles; and in the Derby earthquake of 1903 (Fig. 28), 16J and 8 miles. In many other British earth- quakes, the elongation is equally marked. In one of the Wells earthquakes of 1893, the dimensions are 11^ and 5 miles; in the Exmoor earthquake of 1894, 23 and 12 miles; in one of the Carlisle earthquakes of 1901, 37 and 13 miles; in the Carnarvon XIV] ORIGIN OF TECTONIC EARTHQUAKES 233 earthquake of 1903, 33 1 and 15 miles; and in the Swansea earthquake of 1906, 26 and 14 miles, A few examples of other earthquakes may be given. In the Loeris earthquake (north- east Greece) of 1894. the dimensions of the innermost isoseismal are 17 and 5 miles; in the Constantinople earthquake of the same year, 109 and 24 miles; and, in the Baluchistan earthquake of 1909. 57 and 8 miles. Two explanations of the elongation of the isoseismal lines are possible — one that the vibrations are transmitted with less loss of energy in the direction of their longer axes than in the pev- pendicular direction : the other, that the seismic focus is of con- siderable length and parallel to the longer axes. If the former explanation were correct, the isoseismal lines of an earthquake should be approximately similar in form, and so also should be those of different earthquakes in the same region. But, as will be seen from the ma]) of the Derby earthquake of 1904 (Fig. 51), the innermost isoseismal line may be circular and the next elongated. In the Doncaster earthquake of 1905, the innermost isoseismal consists of two detached circular portions, while the next is elongated. In the series of Inverness earthquakes of 1901 (Fig. 96), the isoseismal lines of some shocks are elongated and of others circular. Lastly, in the Pembroke earthquakes of 1892, the isoseismal lines of some earthquakes are directed north and south, of others east and west. Thus, though the forms of isoseismal lines may be. and are, modified by the nature of the surface-rocks (sects. 45, 46), it seems clear that their elongated forms are chiefly due to the existence of seismic foci of considerable length directed parallel to the longer axes. Moreover, since these axes are almost in- variably parallel or perpendicular to the principal faults of the ei)iccntral district, and since in many cases they are known to lie on the downthrow side of the faults, it follows that the earthquakes must be associated with the faults, most often with the strike-faults, but not seldom with the transverse faults, of the district. 236. Number of Earthquakes greatly in excess of the Number of Faults. Few eartlKpiakes could result from the f()rmati«)n and extension of a fracture, whereas the subsetjuent growth of the fault nuist l)e the result of innunierable slij)s. Now, the number 234 ORIGIN OF TECTONIC EARTHQUAKES [ch. of earthquakes felt in a district is far in excess of the number of faults, and our earthquake-recOrds extend over only a few years, rarely over centuries, whereas the formation of many faults has occupied a large part of geological time. We need only refer to the 143 earthquakes noticed at Comrie, in Perthshire, during the last three months of 1839; to the 306 shocks felt at Zante in the year 1896; or to the 3365 earthquakes recorded at Gifu in Japan, from Oct. 29, 1891, to the end of 1893. In such cases, the earthquakes must be due to the growth of existing faults rather than to the formation of new fractures. 237. Migration of the Epicentre along a Fault. Examples of the migration of the epicentre are given elsewhere in this volume —in the case of the Inverness earthquakes of 1901 in sect. 240 and of the Mino-Owari earthquakes of 1891 in sect. 215. One other instance may be given — that of the Carnarvon earthquakes of June 19, 1903, and 1906, connected with the Aber-Dinlle fault. Denoting slips at the north-eastern end, centre and south- western end by the letters A^, C, S, the distribution of the different epicentres in time is represented as follows, from left to right — N, principal focus, S, N, N, N, N, C, N, N, C, N, C, C, C— N, the last occurring after the laj^se of three years. Now, if the migration were to take place outwards in one or both directions only, this migration might be due to a gradual extension of the fracture: but, since the epicentres retrace their steps, returning to the central region, the corresponding shocks must be referred to fault-slipping rather than to an extension of the fractiu'e. 238. Adequacy of Fault-Growth as a Cause of Earthquakes. It is important to notice that the generation of an earthquake by fault-slipping requires a far less expenditure of energy than that by the formation of new fractures, and this is a matter of consequence seeing that, in all probability, the origin of both must be traced to the slow cooling of the earth. Moreover, in the case of fracturing, the initial disturbance is merely the elastic recoil of rock-particles from the surface of the fracture ; whereas, in the case of slipping, the initial disturbance depends on the mass of the displaced crust, increased in some earthquakes by the sudden motion of the crust. As the weight of the mass XIV] ORIGIN OF TECTONIC EARTHQUAKES 235 may vary from that of a few cubic miles to that of several thousand, or even million, cubic miles, and the displacement from a small fraction of an inch — a mere creep — to several or many yards, it is evident that the friction so generated must be capable of producing earthquakes of every degree of strength — from the slight shocks which are occasionally felt in this country to the destructive earthquakes Avhich visit the greater seismic districts, such as those of Calabria, Central Asia, Chili and Japan. Origin of Simple Earthquakes 239. Earthquakes have been divided, according to the nature of the shock, into the three classes of simple, twin and complex earthquakes (sect. 34). The great majority of slight and moderately strong earthquakes belong to the first class. In Great Britain, for instance, 95 per cent, of the earthquakes are simple, and the remainder twin, earthquakes, the latter being usually of much the greater intensity. The Inverness earthquakes of 1901 may be taken as typical examples of simple earthquakes', and the Derby earthquakes of 1903 and 1904 of twin earthquakes. Complex earthquakes, of which the Mino-Owari earthquake of 1891, the Assam earthquake of 1897, and the Alaskan earth- quakes of 1899, may be regarded as types, will be considered more briefly, as they have already been referred to in Chapter V. 240. Inverness Earthquakes of 1901. The isoseismal lines of the principal earthquake, which occurred on Sep. 18, are repre- sented in Fig. 50; and it has been shown (sect. 129) that the earth- quake was probably due to a slip along the great fault which traverses the whole of Scotland in a south-westerly direction past Inverness, and which is responsible for the marked linearity of its surface-features from Tarbat Ness to Loch Linnhe. Further evidence in sui)p()rt of this connexion is given by the accessory shocks. The boundaries of the disturbed areas of the more important shocks are shown in Fig. 96, and the centres of these areas in Fig. 97. There was first a slight fore- shock (a) at 6.4 p.m. on Sep. 16. Then came the principal earth- quake (/?) at 1.24 a.m. on Sep. 18, the focus extending nearly from Inverness to Loch Ness. At about 1.35 a.m., a slight after-shock (c) originated near the south-west margin of the 236 ORIGIN OF TECTONIC EARTHQUAKES [ch. principal focus. At 3.56 a.m., the strongest after-shock (g) occurred. Its centre was half a mile farther to the north-east, but, as its focus was several miles in length, it must have ex- tended some distance beyond the south-west margin of the principal focus. At 9 a.m., another shock {h) occurred, with its Fig. 96. Map of the principal after-shocks of the Inverness earthquake of 1901. centre half a mile north-east of the principal centre (B), and with a focus slightly overlapping the north-east margin of the principal focus. The next important movement occurred on Sep. 29, with its centre about 1 mile to the south-west of the principal centre. This was followed, on Sep. 30, by one of the XIV] ORIGIN OF TECTONIC EARTHQUAKES 237 strongest after-shocks (/«), the centre of which lay to the south- west of the principal centre, and the focus of which must have extended 2 or 3 miles beneath Loch Ness. Again, on Oct. 18, occurred the last strong after-shock (q), M-ith its focus in the neighbourhood of Dochgarroch. Thus, the foci of the principal earthquake and of all the accessory shocks lie on the downthrow Fig. 97. Distribution of the centres of tiie principal after-shocks of the Inverness earthquake of 1901. side of the fault. It will be noticed also that, in the latter shocks, there is a gradual approach of the centre towards the fault-line, showing that the depth of the corresponding foci gradually decreased with the lapse of time. 241. Origin of Simple Earthquakes. In the case of the In- verness cartlujuakcs, tlure is no e\ idencc to show on which side of the fault the rock was displaced. If, however, the move- ments were a continuation of those which have occurred in the past, it is probable that the principal earthquake was produced by a slight but sudden sag of the crust on the south-east side extending from Inverness to near the end of Loch Ness. In the diagram (Fig. 98), the upper line is supposed to represent the surface of the earth from the Moray Firth (north-east of Inverness) to Loch Ness. The straight dotted line is intended 238 ORIGIN OF TECTONIC EARTHQUAKES [ch. to represent a horizontal straight line traced on the south-east side of the fault-surface through the focus of the principal earth- quake. Owing to the movement which resulted in the earth- quake, this line becomes a curve represented by the continuous line ACB. The distance AB represents the length of the focus, about 8 miles. The distance betweei^ the straight dotted line and the curve at C represents the amount of the subsidence. In the diagram, this distance is greatly exaggerated. In reality, it may have been a fraction of an inch. The first effect of this displacement would be an increase of stress in the terminal regions, A and B, of the principal focus. If the rock were previously near the point of slipping, the additional stress M'ould be sufficient to cause slips in these regions. The effects of these secondary slips would be new in- creases of stress still farther outwards and again in the central Fig. 98. Diagram illustrating the nature of the dis- placement that causes a simple earthquake. region below C, and this would continue until the additional stresses imposed were no longer able to overcome the resistance to movement. The final form of the straight dotted line AB would thus be represented by the curved dotted line A'C'B' . It is evident, also, that a downward movement at the level of the focus would increase the stresses on the fault-surface in the central region above. Thus, there would be a tendency for the foci to diminish gradually in depth. Now, of the six principal after-shocks of the Inverness earth- quake, the first and second occurred at and slightly beyond the south-west margin of the principal focus, the third at and beyond the north-east margin, the fourth near the centre and nearer the surface, the fifth beyond the south-west margin and at a still smaller depth, while the last was in the central region and quite close to the surface*. * C. Davison, Quart. Journ. Geol. Soc, vol.- 58, 1902, pp. 377-397. xiv] ORIGIN OF TECTONIC EARTHQUAKES 239 Origin of Twin Earthquakes 242. Derby Earthquakes of 1903 and 1904. The isoseismal hues of the Derby earthquake of Mar. 24, 1903, are represented in Fig". 28. The two inner isoseismals are elongated in the direction X. 33° E., and they are farther apart on the north-west side. It may be inferred (sect. 129) that the average direction of the fault is N. 33° E., and its hade to the north-west. The fault is probably deep-seated, there being none known in this position. In the greater portion of the disturbed area, the shock con- sisted of two distinct parts separated by a brief interval of rest and quiet. At some places, only one shock was felt. Plotting all such places on the map, they are found to lie within a straight narrow band, about 5 miles wide, that runs centrally across the inner isoseismal lines in the direction W. 34° N., that is, at right angles to the longer axes of the isoseismal lines. This band is known as the synkinetic band. Its boundaries are represented by the broken lines in Fig. 28. Outside this band, and throughout all the rest of the disturbed area, the interval between the two parts was one of rest and quiet. Its average duration was 3 seconds. The isoseismal lines of the earthquake of July 3, 1904, are represented in Fig. 51. In this case, the innermost isoseismal line is a small circle with its centre close to Ashbourne. The sur- rounding curves are elongated in almost the same direction as those of 1903, namely, N, 31° E,, and, as before, they are farther apart on the north-west, than on the south-east, side. The shock again consisted of two distinct parts, except near the boundary of the disturbed area and within a narrow central band. The boundaries of this band cannot be determined with accuracy, but its central line is indicated by the broken line in Fig. 51. This line is curved, its concavity facing the south-west, and it crosses the longer axes of the isoseismal lines at right angles and at a short distance on the north-east side of Ashbourne. The most significant feature of these eartluiuakes is the double nature of the shock. A single impulse in one focus might be duplicated either by underground reflection or refraction, or by the separation of the \ibrations into condensational and dis- tortional waves. The first supposition is inadmissible, owing to 240 ORIGIN OF TECTONIC EARTHQUAKES [ch. the widespread observation of the double shock; the second cannot be entertained, as it would require a continual increase with distance in the interval between the parts of the shock; and, on neither supposition, can the existence of the synkinetic band be explained. For the latter reason, the conception of two impulses in a single focus is negatived. It follows, then, that both earthquakes originated in two distinct foci. One epicentre must be close to Ashbourne, near the centre of the inner iso- seismal of the earthquake of 1904; the other probably about 3 miles west of Wirksworth, and the distance between them about 8 or 9 miles. Now, if the impulses in these two foci occurred simultaneously, the vibrations from both would coalesce along a straight narrow band traversing the disturbed area midway between the two epicentres and at right angles to the line joining them. If, how- ever, the Wirksworth focus were first in action by a second or two, the synkinetic band would be curved, its concavity facing the south-west, for the vibrations from the Wirksworth focus would travel farther than those from the Ashbourne focus before the two series coalesced. If, again, the impulse at the Wirksworth focus had preceded that at the Ashbourne focus by several or many seconds, the vibrations from the former would be felt over all the disturbed area and there would be no synkinetic band. Thus, in the Derby earthquake of 1903, it is clear that the two impulses occurred in the two detached foci at absolutely the same instant. In 1904, the Wirksworth focus was first in action by a second or two. This slight precedence of one impulse was also manifested in the Hereford earthquake of 1896 and the Stafford earthquake of 1916. In the Doncaster earthquake of 1905 and the Swansea earthquake of 1906, there was no trace of a synkinetic band, one focus being in action several seconds before the other. These conclusions are supported by the forms of the isa- coustic lines, which have been drawn for the Hereford earth- quake of 1896 and the Derby earthquakes of 1903 and 1904. Those of the Derby earthquakes are represented by the dotted curves in Figs. 28 and 51. In each earthquake, the isacoustic lines are distorted in the direction of the synkinetic band, the reason being that the sound-vibrations from the two foci were XIV] ORIGIN OF TECTONIC EARTHQUAKES 241 heard simultaneously along and near this band, that they were therefore louder and heard by a larger percentage of observers in this district than elsewhere. Each of the Derby earthquakes was followed by a simple earth- quake of slight intensity — that of 1903 after 40 days (on May 3), and that of 1904 after 8 hours. In each case, the isoseismal axes are approximately parallel to those of the principal earth- quake, and the epicentre lies midway between the two epi- centres of the principal earthquake. The after-shocks must therefore have originated along the same fault as the principal earthquakes, and in the region between the two detached foci of these earthquakes *. s: Fig. 99. Diagram illustrating the nature of the dis- placement that causes a twin earthquake. 243. Origin of Twin Earthquakes. It is possible that the rock along a fault-surface might be on the point of slipping in two detached but neighbouring regions, and that the wa\es resulting from a sudden movement in one region might precipitate a similar movement in the other. But such an explanation is inadmissible for twin earthquakes in which a synkinetic band exists; for, in such earthquakes, the impulses either occur simul- taneously, or the second impulse occurs before the waves from the first focus have time to reach the second. In a simple earth- quake, the displacement which produces it is probably a mere translation of the moving rock. In a twin earthquake, the only method l)y which practically sinuiltaneoiis movements can take place in two detached foci, with little, if any, movement in the interfocal region, is one of rotation about the latter region. * C. Davison, Quart. Juuni. Geol. Soc, vol. (iO, 1904, pj). 215-282; vol. 61, 1905, pp. 8-17. D.M.S. 16 242 ORIGIN OF TECTONIC EARTHQUAKES [ch. The continuous line in Fig. 99 is supposed to represent a section of a thin stratum in a great crust-fold along a fault at right angles to its axis ; A the crest of the fold, S the trough, and M the median limb. Now, if a small step were to take place in the growth of the fold, from the form represented by the continuous line AMS to that represented by the broken line A' MS', there would evidently be two regions of displacement, one between A and A', the other between S and »S", while in the intermediate region of the median limb M, there would be little or no displace- ment. The two displaced regions would thus be the two seismic foci of the twin earthquake, more or less completely detached owing to the almost imperceptible movement of the median limb. If this explanation be correct, the distance between the foci should be approximately the same as that between successive crests and troughs of crust-folds. Both distances are subject to wide variations. The distance between the twin foci of British earthquakes varies from 4 to 23 miles, the average distance being about 10 or 11 miles. For the lengths of British crust-folds, we have no detailed measurements, but the courses of the principal anticlines in France have been mapped, and the average distance between successive anticlines and synclines along several lines lies between 9 and 12 miles. A movement of rotation of a crust-fold, such as that described above, must produce an increase of the stress already existing in the median limb, an increase that, sooner or later, must cause a simple displacement of the limb into a position such as that indicated by the dotted line in Fig. 99. Thus, a twin earthquake should be followed, as in the Derby earthquakes, by a simple earthquake in the region of the fault lying between the two foci of the principal earthquake. Such after-shocks, however, are invariably slight compared with the twin earthquakes themselves. They show that the sub- sequent displacement of the median limb is small compared with that which takes place in the crest and trough of the fold; that, in other words, the crust-fold in its growth becomes accentuated in form more than in its advance along tlie surface of the fault which intersects it*. * JNI. Bertrand, Compi. Rend. Acad. Sci. Paris, vol. 118, 1894, pp. 258- 262; Davison (1), pp. 32-33. XIV] ORIGIN OF TECTONIC EARTHQUAKES 243 244. Deformations of the Crust during Twin Earthquakes. Many earthquakes of great, tliough not of the greatest, intensity are twin earthquakes, as, for instance, theNeapoHtan earthquake of 1857, the Andakisian earthquake of 1884, the Charleston earthquake of 1886, the Riviera earthquake of 1887, and the Messina earthquake of 1908. The warping of the crust during the Messina earthquake has been described in sect. 89. If Omori be correct in his interpretation of the crust-movements observed with the Formosa earthquake of 1906 (sect. 87), it would seem l)robable that this was also a twin earthquake. The fault along which the movement occui-red was a transverse fault, but the structure of the country is somewhat different from that sug- gested in the preceding paragraphs. In Formosa, the fault apparently separates an anticline from a syncline in each half of the fault, the shaded areas in Fig. 41 representing the synclines, and the mishaded areas the anticlines. Origin of Complex Earthquakes 245. The origin of complex earthquakes has already been considered in dealing with the deformations of the crust (Chapter V). The fault-displacements there described were divided into four classes, the last of which is possibly connected with twin earthquakes (sect. 87). In the others, the displace- ments were: (i) mainly horizontal, (ii) partly horizontal and partly vertical, and (iii) mainly vertical. (i) The earthquakes in which the displacement is mainly hori- zontal are apparenth^ connected with strike-faults. The displace- ment usually occurs over a great length of fault (in one case over 290 miles), both sides (in two known cases) move in opposite directions and the amount of the displacement diminishes rapidly as the distance from the fault increases. (ii) The second class consists of the Mino-Owari earthquake of 1 891 . In this case, a displacement occurred along a transverse fault, which crosses most, if not the whole, of the main island of Japan, as well as along a secondary fault without visible disi)laeement at the surface. Apparently, the whole crust in the neighbourhood of the faults was thrust forward in a south- easterly direction, the portion on the south-west side of the main fault advancing farthest. An important difference between 16—2 244 ORIGIN OF TECTONIC EARTHQUAKES [ch. twin earthquakes and complex earthquakes of this class should be noticed. Both are due to movements along transverse faults, but twin earthquakes are connected with the growth of a fold, complex earthquakes are due to the bodilj^ displacement along a great length of the fault. (iii) When the movements are chiefly vertical, they are con- fined to one or several vertical faults or "blatts" or to the upshoot faults of a thrust-plane. Either both sides of the fault move in opposite directions, as in the Alaskan earthquake of 1899, or one (the mountainous) side alone is moved and uplifted, as in the Wellington earthquake of 1855. In both cases, the effect of the uplift is to raise one or more large mountain-blocks and to tilt them slightly in the direction away from the fault or faults. The Assam earthquake, according to Oldham, was caused by a displacement along a thrust-jjlane, 200 miles in length, at least 50 miles in width, and not less than 6000 sq. miles in area — a displacement which involved minor movements along branch-faults and a general crumpling or warping of the surface-crust *. Origin of Accessory Shocks 246. Origin of Fore-Shocks. The origin of accessory shocks is connected with the growing stresses which culminate in a great displacement, and with the various residual stresses which are brought into action by that displacement. The stresses which end in a fault-slip may be the growth of many years, of centuries even. Neither stresses nor the resist- ances opposed to them can be imiform throughout the whole fault-surface. Local obstacles must be overcome before any great general movement can take place, and the removal of these obstacles is effected bj^ local slips, each of which results in a fore-shock. The chief purpose of these slips is thus to equalise the effective resistance to motion over the fault-surface, so that ultimately, when the stresses throughout exceed the resistances, the movement takes place almost instantaneously or with great rapidity over a long expanse of the fault. 247. Origin of After-Shocks. With the displacement which gives rise to the principal earthquake, there at once ensues a * Oldham, pp. 164r-179. XIV] ORIGIN OF TECTONIC EARTHQUAKES 245 change in the stresses to which the neighbouring crust is sub- jected. The sudden increase of stress is reheved by shps — a few considerable, the majority very small — along the fault-surface. At first, the area of this displacement extends outwards ; then it leaves the terminal regions and shrinks continually towards the central regions. If the displacement of the principal earth- quake be partly or mainly an uplift, the weight of the elevated mass aids its return to the position of equilibrium and must be the cause of innumerable after-slips. The after-shocks of such earthquakes should be far niore numerous than those which follow a twin earthquake or an earthquake caused by a displace- ment that is mainly horizontal. In the Concepcion earthquake of 1835 and in the Assam earthquake of 1897, there is some reason for thinking that the nmlti plication of after-shocks may be closely connected with the subsequent fault-movements (sect. 216)*. Origin of Earthquake-Sounds 248. Of the sound-phenomena which accompanj^ earth- quakes, two are of special significance with regard to the origin of the sound. These are the general precedence of the shock by the sound (sect. 73), and the excentricity of the sound-area with reference to the isoseismal lines (sect. 72), As the vibrations which form the sound and shock travel with approximately the same velocity (sect. 73), it is evident, from the precedence of the sound, that the two sets of vibrations originate, in part at any rate, in different regions of the fociis, and that the region from which the early sound-vibrations pro- ceed lies outside the other. The excentricity of the sound-area leads to the same conclusion. It implies that the origin of the soinid- vibrations lies principally in the upper and lateral margins of the seismic focus, for the vibrations from the upper margin would be more readily audible than those, if any, which come from the lower. In the case of a fault-slip, the seismic focus is a siu'face inclined to the horizon. In its simplest form, there is a central region in which the relative disphiccment of the two rock- * This suggestion is supported by the occurrence of the maximum epoch of the diurnal period of after-shocks (sect. 204) at about the same time as that of tlie diurnal period of Imrometric pressure, namely, about midniglit. 246 ORIGIN OF TECTONIC EARTHQUAKES fcH. masses is a maximum, and this is surrounded by a margin in which the relative displacement is small and gradually dies away towards the edges. As the vibrations of great range are also of long period, it is evident that, from all parts of the focus, there start together vibrations of various range and period — the large and slow vibrations from the central region, and the small and rapid vibrations chiefly from its margins. Now, between the sound-vibrations from the margins and the large vibrations from the central region, there can be no dis- continuity of period. Among the vibrations must therefore be included the deepest sound that can be heard by the human ear. It is evident, also, that the intensity of the sound must gradually increase until the shock is felt, after Avhich it must die away. Lastly, the greater strength of the vibrations from the central portion will render audible vibrations of longer period than those which come from the margins, and thus the loud explosive crashes which are heard near the epicentre should accompany the strongest perceptible vibrations. The magnitude of the sound-area depends chiefly on the dimensions of the seismic focus and therefore of its lateral margins. That of the disturbed area depends partly on the size of the focus but chiefly on the initial intensity of the vibrations from its central portion. Thus, with very strong shocks, the sound-area may be a comparatively small district surrounding the epicentre. With slight shocks, the marginal region may be so great compared with the central portion of the focus, that the sound-area may overlap the disturbed area. In the limiting- case, the central portion of the focus will disappear, and a sound will be the only result of the movement that is sensible to human beings. Thus, the earth-sounds, which are so prominent among the after-shocks of a great iearthquake, are merely the repre- sentatives of creeps along the fault-surface — creeps that are so small that they do not give rise to vibrations that can be felt*. 249. Two other points may be referred to here in connexion with the origin of earthquake-sounds. (i) The district represented in Fig. 100 is the central area of the Mino-OAvari earthquake of 1891. The broken lines indicate the boundaries of the strongly-shaken area and the dotted line * C. Davison, Phil. Mag., vol. 49, 1900, pp. 66-70. XIV] ORIGIN OF TECTONIC EARTHQUAKES 247 the course of the fault-scarp. During the years 1885-1892, 3014 earthquakes originated in this district, and 20 per cent, of them were accompanied by sound. The percentage, however, varies throughout the area, and the continuous curves represent this variation. The meaning of the cur\e marked 40 is that, if any point on the curve be regarded as the centre of a small district, then 40 per cent, of the earthquakes originating beneath it were accompanied by sound. Now, as superficial earthquakes would have a greater chance of being heard than deep-seated earthquakes, it follows that the curves of highest percentages in Fig. 100 correspond with the Fig. 100. Distribution of the audible after-shocks of the Miuo-Owari earthcjuake of 1891. earthquakes with the shallowest foci. The axes of the curves thus mark out approximately the lines of growing faults, and show that the displacement which gave rise to the fault-scarp is continued some miles farther to the south-east and that dis- placements also occurred along another fault following the main l)and of the strongly-shaken area (sects. 208. 215). (ii) The Mino-Owari earthquake occurred on Oct. 28, 1891. In the following month, the percentage of audible earthquakes within the area represented in P'ig. 100 was 18, and during the next five months it lay between 10 and 12. Then, in May 1892, it rose suddenly to 39, and during the next seven months never 248 ORIGIN OF TECTONIC EARTHQUAKES [ch. fell below 32, its average being 42, In certain smaller districts, the same change is noticeable. In one, the percentage of audible earthquakes rose from eight during the three months Nov. 1891- Jan. 1892 to 39 during the next eleven months; in another from 10 to 55. Thus, the stresses produced by a fault-slip are increased in the portions of the fault adjoining the focus and especially in that above it. By slip after slip, the stresses are gradually re- lieved, until, even at the surface, they are no longer capable of producing the minute creeps which are perceptible to us as earth-sounds *. Conclusion 250. According to the theory described in this chapter, earth- quakes are merely the passing signs of the changes which are now taking place in the earth's crust. The districts in which earthquakes are most numerous and violent are those in which the crust-changes are being effected most rapidly. In the history of the earth's crust, the period over which our seismic records extend is infinitesimal. During that period, some regions have been almost quiescent while others have been fre- quently shaken. It does not follow that, if the period could be sufficiently prolonged, the conditions of the two regions might not be interchanged. So far as our evidence goes, however, the distribution of seismic activity considered in Chapter X represents the portions of the earth's crust that are now growing. There are but few parts of the globe to which the term "aseismic" can be strictly applied. Even in Great Britain, there are ancient faults which are yet in a state bordering on motion, and crust-folds that are still being intensified. When we turn to moiuitainous districts of more recent growth, we find the same movements taking place, but of far greater strength and frequency. In Central Asia, some of the moiuitain- ranges are growing by leaps and bomids. Farther south, the Himalayan masses, as we see from the Kangra earthquake of 1905, are still being forced over the fringe of Tertiary mountains which separate them from the plains of India. * C. Davison, Phil. Mag., vol. 49, 1900, pp. 50-52. XIV] ORIGIN OF TECTONIC EARTHQUAKES 249 In other districts, there are mountain-ranges in an earlier stage of groAv-th. The western boundary of the Pacific Ocean is the most unstable region in the globe. In its festoon-shaped groups of islands, of which the Japanese Empire is typical, the crust is being pressed over the "fore-deeps" be5^ond just as in the older ranges of the Himalayas. In both regions, this mode of growth results in the steeply-sloping surface which is typical of our principal seismic regions. i\.s to the precise cause of the great and widespread move- ments, we are still ignorant. The cause may be one that resides entirely within the earth. But, when we consider the close coincidence of disturbances in regions so remote as the western and eastern boundaries of the Pacific Ocean, when we trace the connexion which apparently exists between the greatest earth- quakes and the small migrations of the Pole, we realise that it is not impossible that we maj^ have to look beyond our globe and recognise that other bodies of the solar system may claim a share, not only in the movements of the earth, but also in the oTowth of its surface-features. INDEX Accessory shocks, 4, 199, 244 After-shocks, 5; diurnal periodicity, 195: number, 204; intensity, 205; decHne in frequency, 206; in in- tensity. 208 ; focal migration, 210; connexion with the nature of the principal earthquake, 211; and the height of the fault-scarp, 212; origin, 19G, 2+4 Agamennone, G., 28, 145. 208, 224; microseismometrograph, 28 Alaskan earthquakes of 1899, 69-74, 77. 79, 81. 83, 85. 92, 93, 100, 114, 174, 232, 235, 244 Alban Hills earthquakes, 223, 225, 226 Alippi. T.. 56, 64, 65 Andalusian earthquake of 1884, 106, 129. 141, 174, 243 Annual seismic period, 183, 190; origin, 191 ; distribution of maxi- mum-epochs in Japan, 192 Arcidiacono, S.. 131, 132, 221 Asama-vama (Japan) earthquakes, 127, 221 Asia Minor earthquake of 1909, 142 Assam earthquake of 1897, 5, 36, 41, 50, 69-71, 74, 79-81, 83, 89, 101- 105, 107-109, 111-113, 117, 145, 200, 204, 205, 208, 212, 235, 244, 245 Avalanches, 101, 102 Baluchistan earthquake of 1892, 70, 75. 77. 78, 81, 82. 105, 232; of 1909, 233 Baratta, M., 42, 65, 141, 162. 174, 220. 221, 223, 224 Barisal guns, 63 Bengal earthquake of 1885, 50, 129, 131 Bertrand. M., 242 Bluff-fissures. 106 Bolton earthquake of 1889. 66, 125 Bonin Islands carth(|uakc of 1914, 142, 148. 1.54 Bovs, C. v., 141 Brfgham, W. T., 63, 190 British Association Seismological Conunittcc, 22, 134, 135, 141 , 167, 178 British earthquakes, nature, 32; disturbed area, 49; forms of iso- seismal lines, 49, 232; sound- phenomena, 56-59, 61, 67 ; annual periodicity, 190 ; after-shocks, 212 ; migration of epicentres, 234; simple earthquakes, 235; twin earthquakes, 239 Brontides, 63 ; nature, 63 ; frequency, 64; origin, 64; in Italy, 63; in the Philijjpine Islands, 64, 65; in Haiti, 65; in Belgium, 65 Cachar earthquake of 1869, 50, 129 Calabrian earthquakes of 1783, 102, 173; of 1905, 173 Californian earthquake of 1906, 5, 32, 41, 42, 50. 69. 70, 72, 75, 77- 79, 81, 82, 100, 101, 103-105, 111, 112. 142, 145, 156, 174, 200, 213, 231, 232 Cancani, A., 28, 47, 56, 64, 65, 198; intensity scale, 47; microseismo- metrograph, 28 Carnarvon earthquake of 1903, 212, 232, 234 Cavasino, A., 131, 132, 208, 213 Centrum, 3 Charleston earthquake of 1886, 5, 32,48-50, 101, 105, 106, 111, 112, 132, 161, 199, 212, 243 Close, M., 12 Clustering of great earthquakes, 179 Colchester earthquake of 1884, 109, 111 Complex earthquakes, nature, 32; origin, 243 Compression of alluvium, 104 Comrie earthquake of 1839, 62, 213, 234 Concepcion earthquake of 1835, 92, 212. 245 Condensational waves, 134 Constantinoj)lc eartlujuake of 1894, 132, 233 Damping of seismographs, 15 Darwin. C, 213 Darwin, G. H., 191 Darwin. H., 7, 12, 17. 22; bihlar pendulum, 15, 17, 22 Davison, C, 5, 42. 43. 53. 56-63, 67. 68. 76-, 77. 79. 98. 99. 105. 124, 1.56. 163. 183, 186. 190. 194. 196, 199. 202, 203. 211. 213. 214. 21(), 229, 230, 238, 241, 242, 246, 248; sound-scale, 57, 63 252 INDEX Deformations of the earth's crust, nature, 69; measured by trigono- metrical re-surveys, 82, 83 Depth of seismic focus, determina- tion of, 125; methods depending on time, 125; on direction, 128; on intensity, 130 Depth, relative, of seismic foci, 132 De Ranee, C. E., 110 Derby earthquake of 1903, 15, 43. 49," 60, 212, 232, 235, 239, 240, 241 ; of 1904. 49, 53, 122, 212, 233, 235, 239, 240, 241 Direction of earthquake-motion, 50, 119 Distortional waves, 134 Distribution, laws of seismic, 168; connexion with surface-gradient, 168; with secular changes of ele- vation, 172 Disturbed area, 4; magnitude, 49 Diurnal seismic period, 183, 194; of after-shocks, 195 Doncaster earthquake of 1905, 59, 122, 212, 233, 240 Double earthquakes, 180 Drummond, J., 62 Duration of earthquakes, 54; of pre- liminarv tremor, 53 Dutton. C. E., 5, 36, 49, 101, 105- 107, 111, 112, 130-132 Earth-flows, 101, 104 Earth-lurches, 104 Earth-pulsations, 3 Earthquake, definition of, 1 Earthquake-motion, nature, 7; per- sonal impressions, 30; seismo- rn^aphic evidence, 32; perio.d of vibrations, 37 ; maximum accelera- tion of vibrations, 38, 39, 41 ; range of vibrations, 37; direction, 50; duration of preliminary tremor, 53; duration of earthquake, 54 Earthquake-sound, 3 Earth-slumps, 101, 103 Earth-sounds, 3; in Meleda Island (Adriatic), 62; at East Haddam (Connecticut), 62; at Guanaxuato (Mexico), 63 Earth- tremors, 3 Earth's interior, nature, 156 East Haddam (Connecticut) earth- sounds at, 62 Ecuador earthquake of 1906. 93, 96, 98 Eginitis, D., 131. 132 Ehlert, R., 6 Elevation, secular changes and earthquake frequency, 172 Emergence, angle of, 3 End portion of earthquake-motion, 32 Epicentre, 4; determination of, 116, 157; methods depending on time, 118; on direction, 119; on inten- sity, 121; distance of, 148 Etnean earthquakes, 133, 217, 225 Ewing, J. A., 6, 11, 17, 19-22, 32; duplex pendulum seismograph, 21, 36; seismograph, 17 Exmoor earthquake of 1894, 232 Fault-displacements during earth- quakes, 70; general appearance. 71; length, 74; form, 75; relation to structure of ground, 76; hori- zontal displacements, 77, 82; vertical displacements, 79, 83; horizontal and vertical displace- ments, 84; complex displace- ments, 84; probable connexion with warping, 89 Faults, earthquakes and the growth of, 230, 234 Ferruzzano (Italy) earthquake of 1907, 99 Fissures, nature, 105; classification, 106; bluff-fissures, 106; fissures unconnected with excavations, 107: hill-foot fissures, 108; fault- block fissures, 109 Fieri (Etna) earthquake of 1914, 226 Focus, seismic, 3 Fondo Macchia (Etna) earthquake of 1865, 217, 225; of 1911, 217 Forbes, J. D., 28, 29; inverted pen- dulum, 28 Forel, F. A., 45, 48 Fore-shocks, 5: number and inten- sity, 199; distribution of Mino- Owari fore-shocks in time, 200; in space, 202; prevision by means of fore-shocks of great earth- quakes, 202; origin, 244 Formosa earthquake of 1906, 69, 70, 75,76,78,79,81,84 Frequency of earthquakes, relation with intensity, 161, 178; annual frequency, 177; synchronous vari- ations, 181; secular variations, 182 Fuchs, C. W. C, 190 Fuller, M. L., 69, 89, 101, 102, 108, 109, 111, 113 Galitzin. Prince B.. 6, 12, 14, 15, 17, 22, 27, 127. 156, 160; seismo- graph, 14, 15, 17, 22, 27 Geiger, L., 134, 146 INDEX 253 Geinitz, F. E., 90 German, mid-, earthquake of 1872, 126; south, earthquake of lUll, 127 Glaciers, effects of earthquakes on. 114 Gouffres, 63 Gray, T., 11, 12, 18, 20, 21, 27, 82. 54, .55 Gray-Milne seismograph, 20 Guanaxuato (Mexico), earth-soinids at, 63 Guatemala earthquake of 1902. 174, 180 Gutenberg, B., 134 Haug, E., 166 Helston earthquake of 1898, 49, 67 Hengeller, I... 12. 27 Hereford eartiiquake of 1896. 42, 43, 48, .53. 58, 59, 61, 200, 232 Herzogenrath earthquakes of 1873 and 1877. 126 Hirata, K., 37, 51, 55, 129 Hobbs, W. H., 69, 75, 76, 79. 80 Hochstetter, F. von, 90 Hogben. G.. 125. 190 Hokkaido earthquake of 1894, 195, 204. 208 Holden, E. .S., 190 Honda, K., 90, 93, 94, 96-98 Horizontal-motion seismographs, 17 Humboldt, .A. von, 63 Hypocentre, 3 Intensity of shock. 4; lower limit of sensible motion, 38: intensity and nature of groimd, 41 ; in j)its and on surface. 43; scales of in- tensity, 44; relation with fre- quency of earthquakes, 161 Internal reflection of earthquake- waves, 153 International Seismological Associa- tion, 141. 178 Inverness cartluiuake of 1901. 59, 121. 124. 125. 212. 232-235. 240 Iquique eartlujuake of 1877, 90, 92- 95. 97. 99, 141 Isacoustic lines. 4: construction. 59; forms in twin eartlu|uakes. 240 Ischian eartlupiakes, 129, 222, 225. 226 Isitani. I).. 90 Isoseismal lines, 4; construction, 48; forms, 49, 232; determination of epicentre by, 121 ; cause of elon- gated forms, 233 Issel, A., 43 Italian earthquake of 1873, 61 Italian earthquakes, brontides, 63; crust-displacements, 88, 219; sea- waves, 91; depth of foci, 129; volcanic earthquakes, 217, 225 Japan, seismic maps of, 163, 169, 170 Japanese earthquake of 1707, 171; of 1833. 171 ; of 1854, 92. 98, 171, 199. 205: of 1872, 171: of 1887, 33, 37, .50; of 1895, 35 Japanese earthquakes, nature of shock, 33. 37 ; direction of motion, 50; duration, 53; fault-displace- ments. 74-79, 86; sea-waves, 92; distribution in space. 163; migra- tion of epicentres, 173; frequency, 178: annual periodicity, 190; di- urnal periodicity, 194; accessory shocks, 200, 247; sympathetic shocks, 213 ; volcanic earthquakes, 221 Johnston-Lavis, H. J., 129, 132, 223 Jones, E. J., 129 Kacoshima earthquake of 1893, 204 Kangra eartiiquake of 1905, 31, 43, 70. 71, 87, 122, 123, 132, 142, 145, 203, 248 Kashmir earthquake of 1885, 129 Kikuchi. D.. 97, 100 King, F. H.. 110 Kingston earthquake of 1907. 144. 205 Klotz, O., 147, 150. 1.59 Knott. C. G.. 6. 12. 134, 136-140, 150-153, 156, 157, 183, 186, 190, 198 Koto, B., 69, 75, 77, 79, 80, 86 Krakatoa sea-waves of 1883, 90, 93, 96, 98. 99 Kumamoto earthquake of 1889, 141, 204, 208 Kusakabe, S., 135 Lais, R.. 178 Lamb. H., 1.50 Landslips. 101 ; condition of forma- tion. 102; effects on water- courses. 103 Lasaulx. .\. von, 126 Latitude, variation of, and fre- quency of earthquakes, 196 Lawson,'A. C, 5, 7, 69, 75. 77, 79, 80. 83. 101, 104, 105, 107, 111, 112, 213 Lebour. G. A., 230 Linera (Etna) earthquakes of 1914, 218, 226, 227 254 INDEX Lisbon earthquake of 1755, 92, 99, 141, 161, 162, 174 Locris (Greece) earthquakes of 1894, 70, 72, 75, 77, 81, 204, 233 Long waves, 143, 150; returns of, 155 Lyell, C, 69, 75, 77, 80 Macroseisni, 3 Mallet, R., 3, 4, 90, 99, 102, 107, 108, 111, 112, 116, 119, 120, 121, 128- 131, 162, 164. 165, 168, 177, 190, 193, 215, 216 Malvern earthquake of 1907, 125 Maps, seismic, construction, 162; mapping of disturbed areas, 162; of meizoseismal areas, 162; of epicentres, 162 Marinas, 63 Marr, J. E., 80 Marsica (Italv) earthquake of 1904, 132; of 1915, 132, 145 Martin. L., 69, 75, 80, 85, 93, 100, 115 Marvin, C. F., 6, 11, 15, 22, 28; in- verted pendulum, 22, 28 McAdie, A., 47; intensity scale, 47 McMahon, A. H., 75 McMahon, C. A., 75 Megaseism, 3 Meizoseismal area, 4 Meldola, R., 110, 111 Meleda Island (Adriatic), earth- sounds at, 62 Mercalh, G., 45, 46, 129, 216, 223; intensity scale, 45, 46 Messina earthquake of 1908, 32, 41, 42, 69-71, 87. 88, 90, 92, 93, 96, 99, 100, 156, 200, 205, 208, 212, 243 Michell, J., 119 Microseism, 3 Micro-tremors, 3 Middlemiss, C. S., 31, 43, 88, 107, 129, 131, 132 Migrations of seismic activity, 173 ; small migrations in Calabria, 173; in Japan, 174; large migrations, 174; along a fault, 234, 235 Milne, D., 190 Milne, J., 12, 14, 17, 20-25. 32, 40, 42-44, 54, 55, 79, 90, 99, 102, 103, 105, 107. 129. 134, 141, 145-147, 158, 159, 163, 167, 169, 171, 172, 176-180, 182, 190, 196-198, 206, 215, 216: seismograph, 14, 22, 23, 142 Milne-Shaw seismograph, 14, 15, 17, 23, 24, 142 Mining districts, earth-shakes in, 228 Mino-Owari earthquake of 1891, 37- 39, 41, 43, 49, 50, 52, .53, 62, 70, 72-79, 81, 84, 100, 102, 103, 117, 173, 195, 199. 200, 202, 204-206, 208-214, 232, 234, 235, 243. 246, 247 Mist-poeffeurs, 63 Monastir earthquake of 1911, 160 Montessus de Ballore, F. de, 161- 163, 165-167, 169, 171, 177, 178 Mouchketow. J. V., 129 Multiple earthquakes, 180 Nagaoka, H., 135 Nature of ground and intensity of shock, 41 Neapolitan earthquake of 1857, 5, 116. 119-121, 128, 129, 243 New Madrid earthquakes of 1812, 69-71, 87, 88, 101, 102, 107-109, 111, 113 New Zealand earthquakes, depth of focus, 125 Nicolosi (Etna) earthquake of 1901, 225 Ochil earthquakes of 1900-1914, 125 Oddone, E., 131, 132 Oldham, R. D., 5, 36, 41, 69, 75, 80, 81, 89, 101-103, 105, 107-109, 111, 112, 114, 117, 118, 134, 145, 148, 156, 157, 205, 206, 213. 244 Oldham, T.. Ill, 112, 129 Omori, F., 12, 22, 26, 27, 36-43, 47, 51-55, 62, 69, 75, 76, 79, 80, 86, 87, 93, 96, 97, 118, 119, 127, 129, 132, 134, 145, 156, 170, 171, 178, 180, 192-194, 196-199, 205-208, 217. 221, 222, 225, 243; horizontal pendulum. 22, 25; intensity-scale, 47. 49 OReilly, J. P., 163 Origin of accessory shocks, 244; after-shocks, 244; complex earth- quakes, 243; earthquake-sounds, 245; fore-shocks, 244; simple earthquakes, 235, 237; tectonic earthquakes, 230, 248 ; twin earth- quakes, 239, 241 ; volcanic earth- quakes, 227, 229 Overlapping means, method of. 186 Owens Vallev (California) earth- quake of 1872, 69. 70, 75, 76, 78, 81, 83 Papavasiliou, S. A., 75, 77 Partsch, P., 62, 63 Patras (Greece) earthquake of 1889, 141 INDEX 255 Pembroke earthquake of 1892, 212, 233; of 1893. 212 Pendleton earth-shake of 1905, 228 Periodicity of earthquakes, 182; graphical representation, 183; method of analysis, 185; annual period, 183, 190; diurnal period, 19-i; of after-shocks, 195; in rela- tion to intensity, 193 Perrev, A.. 177. 190 Perro't, A., 12 Peruvian earthquake of 1868, 90, 93, 98 Phases of earthquake-motion, 30, 32. 142 Pits, earthquake-motion in, 43 Platania, G., 90, 92, 97, 99, 220, 229 Preliminary tremors, 30, 32, 53, 142 Prevision of earthquakes, 202 Primary waves, 142, 148, 152 Principal portion of earthquake- motion, 32 Principal shock, 4 Propagation of earthquake-waves to great distances, 134; early obser- vations, 141 ; examples of earth- quake-records, 142; phases of motion, 142 Hailway-lines, buckling of, 105 Rayieigh, Lord, 150, 154 Hays, forms of seismic, 150 Hebeur-Paschwitz, E. von, 12, 141 Recording apparatus of seismo- graphs, 12 Reflection of earthquake-waves, 135, 1 53 Refraction of earthquake-waves, 135 Reid. H. F., 6, 75, 83, 135, 145, 203 Rhenish earthquake of 1846, 126 Ricco. A., 220 Ripples in earthquake-motion, 32 River-channels, rise of, 113 Riviera earthquake of 1887, 43, 50. 91. 129, 141, 200, 205, 208, 212, 243 Rossi, M. S. de, 45, 48 Rossi-P^orel intensity scale, 45, 48, 49 Roumanian earthquake of 1893, 141 Rudolph. E., 175, 176; scale of sea- quake intensity, 175 Sadcrra .Maso, M., (i5 Saknra-jinia earthquake of 1914, 216, 225 Sand-blows, 113 Sand-craters, 111 Sand-sloughs, 113 Sand-vents, 111 Sanriku (.Japan) earthquake of 1896, 92-94, 98, 169 Scherer, J., 65 Scheu, E., 178 Schmidt, A., 127 Schmidt, J. F. J., 190 Schuster, A., 183 Serope, G. P., 217 Seaquakes, 1, 139, 175 Sea-waves, seismic, 4; classification, 90; frequency. 91; nature near epicentre, 91; height, 92; effects of, 93 ; distances traversed by, 93 ; nature at great distances, 93; mean velocity, 98; connexion with mean depth of ocean, 98; origin, 99 Secondary effects of earthquakes, 101 Secondary waves, 143. 148. 152 See, T. J.' J., 230 Seebach, K. von, 126, 127 Seiches, seismic, nature, 94; periods, 96; cause of periodicity, 97 Seism, 3 Seismic countries, classification, 161 Seismic focus, determination of form and position, 123; of depth, 124 Seismic maps, construction, 162; of Japan, 163: of the world, 164; Mallet's, 164; de Montcssus dc Ballore's, 165; Milnes, 167 Seismograph, Agamennone. 28 ; Can- cani, 28; Darwin bifilar pendu- lum, 15,17,22; Ewinsr, 17, 21, 36; Forbes, 28; Galitzin. 14, 15, 17, 22, 27; Grav-Milne, 20; Marvin, 22. 28; Milne, 14. 22, 23; Milne- Shaw, 14, 15, 17, 23. 24; Omori, 22, 25; Vicentini, 22, 27; Wie- chert, 15, 17, 22, 29 Seismographs, object, 6; essential parts, 8 ; steady mass. 8 ; supports, 11; recording apj)aratus, 12; test of accuracy, 12; damping, 15 Seismology, scope of, 1 Seismoscope, 6 Sekiya, S.. 36, 37, 43, 51. 179 .Semi-annual seismic period. 191 Shaw, J. .1., 14, 17, 22, 24, 25 SluK-k, definition of earthquake-, 3 Sillcin cartlupiakc of 1858, 126 Simple cartli(|uakes, nature, 31; origin, 235 Snow-avalanches, 114 Sonora (U.S.A.) earthquake of 1887, 70, 81 256 INDEX Sound-area, 4; and distvubed area, relative position, 65 Sound-phenomena, 3; general char- acter of sound, 56; types, 57; in- audibility, 58; variation in audi- bility throughout sound-area , 59 ; in character, 60 ; relative magni- tude of sound-area and disturbed area, 61; relative position, 65; time-relations of sound and shock, 67; vertical arrival of sound- waves, 140 ; origin of earthquake- sounds, 245 Sound-scale, Davison, 57 Springs, effects of earthquakes on, 109 Srimangal (Assam) earthquake of 1918, 132 Steady mass, theory of, 8 Stuart, M., 131, 132 Submarine earthquakes, distribu- tion of, 175 Sumatra earthquake of 1892, 70, 71 , 75, 81, 83 Supports of seismographs, 11 Sussultatory shocks, 31 Swansea earthquake of 1906, 212, 233, 240 Symons, G. J., 90 Sympathetic shocks, 5, 213 Synkinetic band, 239 Syracuse earthquake of 1895, 132 Taramelli, T., 129 Tarr, R. S., 69, 75, 80, 85, 93, 100, 115 Tectonic earthquakes, 5; origin, 230 Teleseism, 3 Tenpo (Japan) earthquake of 1830, 204 Terada, T., 90 Time-curves of earthquake-phases, 144 Tokyo earthquake of 1889, 141 ; of 1894, 34, 35, 37, 50, 51, 119 Tsukuba, Mount (Japan), earth- quakes, 35, 39, 55 Tsunami, 4, 90 Turner, H. H., 135,146-148, 151,216 Twin earthquakes, nature, 32; origin, 239, 241; and deformation of crust, 243 Underground water, effects of earth- quakes on, 109; on springs and wells, 109; extrusion of water from fissures, 110 Undulations, slow, 33 Undulatory shocks, 31 Unfelt earthquakes, early observa- tions, 141 Usu-san (Japan) earthquakes, 222, 227 Valparaiso earthquake of 1906, 92, 93, 96, 98, 174, 180 Van den Broeck, E., 56 Velocity of earthquake-waves, 147 Verny (Turkestan) earthquake of 1887, 61, 129; of 1889, 141 Vertical motion in earthquakes, 33 Vertical-motion seismographs, 17 Vibrations, elements of, 1 Vicentini, G., 22, 27; microseismo- graph, 22 Visible undulations, 36 Volcanic earthquakes, 5, 215; rela- tions with tectonic earthquakes, 215; earthquakes of active vol- canoes, 217; Etnean earthquakes, 217; Japanese volcanic earth- quakes, 221, 224; earthquakes of dormant and extinct volcanoes, 222; characteristics, 224; small depth of foci, 226; origin, 227, 229 Vorticose shocks, 31 Walker, G. W., 6, 135, 149, 160 Warping of earth's crust during earthquakes, 71, 87; general warping, 87; local warping, 88; probable connexion with faulting, 89 Water, extrusion of, from fissures, 110 Wellington (N.Z.) earthquake of 1855, 70, 75, 77, 79, 81, 83, 84, 100, 244 Wells earthquake of 1893, 232 ■ Wells, effects of earthquakes on, 109 West, C. D., 40; formula for maxi- mum acceleration, 40 Wharton, W. J. L., 90, 98, 99 White, W., 110, 111 Wiechert, E., 6, 17, 22, 29, 135, 146, 154, 156; inverted pendulum, 15, 17, 22, 29 Wood, H. O., 42 Yokohama earthquake of 1880, 129 Yoshida, Y., 90 Zeissig, G., 159 Zenkoji (Japan) earthquake of 1847, 204 Zoeppritz, K., 135 Zollner, F., 12, 27 CAMBRIDGE: PRINTED BY J. B. PEACE, M.A., AT THE UNIVERSITY PRESS 331a D299m Daviscn - A manual o^ seismology. ENGINEERING AND MATHEMATICAL SCIENCES LIBRARY University of California, Los Angeles Please return to the above library NOT LATER THAN DUE DATE stamped below. JOl 28 R^^ 30m-8,'66(G493l84)0393 '-^Krr, Engineering & Mathematical Sciences Library 0£ AUXILIARV ?TACK JaL72