Bilham, R., Location and magnitude of the 1833 Nepal earthquake and its relation to the rupture zones of contiguous great Himalayan earthquakes, Current Science, 69(2), 155-187, 25 July 1995.

Summary. One of several pre 20^th century earthquakes in the central gap of the Himalaya occurred in 1833. Prior to this 1995 study its location and magnitude were not well known. Newspaper accounts and official documents were searched for information on the felt effects of the event and a magnitude of M7.7 has been assigned to it. Its location is apparently in northern central Nepal. It was not large enough to release significant accumulated slip from Asian/Indian convergence. Since publication, the location of the garrison town of Mullye has been identified on Thomson's New General Atlas of 1817. Mullya lies due south of Kathmandu at 27°N, 85°E, approximately 15 km NE of Bettiah. It is closeto the location of the current Nepal/India border town of Raxaul.

Location and Magnitude of the 1833 Nepal Earthquake and its Relation to the Rupture zones of Contiguous Great Himalayan earthquakes

Roger Bilham

CIRES and University of Colorado, Boulder CO, 80309-0216

Abstract

Accounts of an earthquake on 26 August 1833 which was felt over a large part of northern India have been interpreted by some authors to represent a great Himalayan thrust event beneath Western Nepal. However, details of the event in the Indian press of 1833 and scientific journals of that time, suggest that the epicenter of the earthquake was near Kathmandu within, or close to, the inferred rupture zone of the Bihar 1934 earthquake. Estimates of moment magnitude based on reported intensities indicate that the earthquake was 7.5<M<7.9, and as such may have done little to release elastic strain accumulating in the region of the Central Himalayan seismic gap, contrary to the expectation of some authors. The location of the epicenter was probably N or NE of Kathmandu, adjoining or overlapping the rupture area of the great 1934 Bihar/Nepal earthquake. The Moment Magnitudes of great Himalayan earthquakes in 1897, 1905, 1934 and 1950, and smaller recent events are compared using recently published empirical relations between isoseismal areas and moment magnitude (Johnston, 1994). When due allowance is made for deficiences in field data, reasonable fits are obtained for all events except for the 1905 Kangra earthquake. The intensity VIII area for this event is anomalously small for an M≥8 earthquake associated with several meters of slip. It is proposed that the Kangra earthquake may have been a slow earthquake.

Introduction

Approximately half of the Himalayan arc has ruptured in four great earthquakes in the past 100 years (Figure 1). The largest region between the rupture zones of these recent events is a 500-800 km segment of the Himalaya between the 1905 Kangra and the 1934 Bihar earthquakes, approximately between the longitudes of Kathmandu and Delhi. Of importance in estimating the present slip potential of this segment, termed the Central Gap by Khattri and Tyagi (1983), is the existence and severity of great historic earthquakes that may have ruptured all or part of the gap. A severe earthquake occurred in Nepal in 1255 when "innumerable towns were utterly destroyed and thousands of their inhabitants killed" (Campbell, December 1833) but the regional extent of this event is unknown. Other large pre-XX century earthquakes in Nepal (1408, 1681, 1810, 1833, and 1866) are mentioned by Chitrakar and Pandey (1986) but none appear to have been as damaging as the 13th century event, causing concern that considerable elastic strain may be available presently to drive one or several M>8 earthquakes in the Central Gap.

An alternative mechanism to absorb slip between Tibet and India is to invoke the possibility of aseismic slip (slow earthquakes or creep) over at least part of the region. Leveling data and recent GPS measurements between India and central Nepal (Jackson and Bilham, 1994; Bürgmann et al., 1994) suggest that creep processes that might otherwise release Indo-Asian convergence aseismically have been insignificant in the past few years. If similar creep rates (2.5±2.5 mm/year) exist elsewhere along the arc throughout the seismic cycle they are evidently inadequate to accommodate completely the slip budget between India and southern Tibet, although they may delay rupture (Bilham et al., 1995). The possibility that some Himalayan earthquakes may be slow events, with large slip but little radiated high-frequency seismic energy, cannot be excluded (Sacks and Linde, 1981; Beroza and Jordan, 1990). Such events would not appear in the historical record as great earthquakes although they could, in principal, release the elastic strain associated with one.

Earthquakes in 1803, 1833 and 1866 appear to have occurred at least partly within the central gap (Khattri, 1987) and the largest of these in terms of felt area is believed to be the 1833 event. Reports of the 1833 earthquake are found in newspapers starting the day after the earthquake, and these and other data are collated in three issues of the Journal of the Asiatic Society of Bengal in the months following the earthquake by Prinsep (1833) and Campbell (1833), and by Baird Smith in two articles a decade later (1843,1844). Summaries of these summaries are found subsequently in various catalogs and comparative studies: Mallet 1852, 1855; T. Oldham 1883; R. D. Oldham, 1897; Dunn et al. 1939; Bapat et al, 1983 and Dunbar et al. 1990. The compilation by Dunbar et al. 1990, lists the event as severe and records its location as 25.1 N and 85.3 E , near Patna south of the River Ganges, at the southern limit of intense shaking described in 1833 reports. A location west of Kathmandu is favored by some authors (Seeber and Armbruster, 1981) who suggest tentatively that it may have occurred in the Central Himalayan Gap. Khattri and Tyagi (1983) place the earthquake approximately 130 km west of the Bihar 1934 epicenter on the edges of the Central Gap and assign the event M=7.6, a location and magnitude consistent with the findings of the present study. One purpose of this article is to estimate more precisely the location and magnitude of the 1833 Nepal earthquake using authentic accounts found in newspapers and scientific articles published soon after its occurrence. A second purpose of the article is to reconcile the intensity data from the 1833 earthquake with the felt reports of four great Himalayan earthquakes and other earthquakes that have occurred since then.

The 1833 earthquake

Just before midnight on 26 August 1833 (23:35 Calcutta time) a 1 million km² region of northern India, Nepal and Tibet was shaken by a strong earthquake which, triggered landslides and rockfalls, destroyed more than 4600 dwellings and many temples, but apparently resulted in fewer than 500 fatalities. It is certain that the loss of life would have been far more severe had not the mainshock been preceded by two large foreshocks five hours and 15 minutes before the mainshock that drove people outdoors in alarm. In some villages in northern Nepal and southern Tibet 30% or more houses were destroyed. Damage in India was less severe with fewer than 10 reported fatalities and few buildings totally destroyed.

The mainshock was felt from Delhi to Chittagong (Fig. 1). It was not felt in Lhasa nor are there reports of the shock further SSW than Jabalpur. Intensities were high in the mountainous region near and north of Kathmandu, and reduced rapidly to the south and more slowly to the east and west. Intensities remain high at the northern limit of high intensity observations in southern Tibet. Accounts of damage where shaking was most intense suggest a similar intensity distribution to that observed during the Bihar 1934 earthquake with the principal exception that accounts of liquefaction features in 1833 are rare. Passes to Tibet were blocked by landslides, and the Kamla River was dammed by a landslide that burst 4 days after the event flooding the village of Baldeah near the India/Nepal border (Bengal Hurkaru, 16 Sept. 1833).

Fig. 1 Approximate Modified Mercalli Intensity VIII zones for 5 large earthquakes in the Himalaya since 1833 and settlements reporting information concerning the 1833 shock (white dots). A great earthquake may not have occurred in the 800 km segment of the Himalaya between the 1905 Kangra and 1934 Bihar earthquakes (Central Gap) for 650 years.

Isoseismal Coverage and Accuracy

The fidelity with which Indian newspapers reported effects of the 1833 earthquake can be gauged by the faithful reprinting of articles printed in one by another. More than a dozen newspapers, and weekly and monthly journals were published in Calcutta (The Reformer, Calcutta, 3, 141, Oct. 27, 1833) although not all of them reported news of the event, and a few are no longer available. Newspapers in Delhi, Bombay, Meerut and Madras also contain accounts of the earthquake. In the 52 accounts in the Appendix, 97 towns and villages are mentioned. Most of these reports can be assigned Modified Mercalli Intensities V-IX with reasonable certainty. Few “barely felt” reports are available and fewer “not felt” reports (e.g. Nazirabad, Xigatse and Lhasa), and the general sparcity of intensity IV or smaller observations prevents reliable estimates of those areas that experienced the earthquake weakly. The absence of details of construction methods of those houses destroyed, or of those surviving, means that intensity estimates are not as reliable as those for the 1934 earthquake (Dunn et al. 1939), and although it is likely that construction practices changed little between the two earthquakes, it is certain that urban dwelling units and civic buildings were less numerous in 1833. This may influence comparisons between the 1833 and 1934 earthquakes because several damage reports are from urban regions near factories or administrative centers for both earthquakes. The Gangetic Plain except for the Kingdom of Oudh, south of western Nepal, was under British colonial rule in 1833 (Walker, 1833). However, reports from Oudh are available despite a significant cholera epidemic in Lucknow and Allahabad. With perhaps the single exception of data from Monghyr (Monger) most of the reports contributed to Indian newspapers are in the form of personal accounts of the earthquake with no statistical descriptions of urban damage. Few fatalities are reported: three during the mainshock at Chapra and one for an aftershock in Bhagalpur. In some accounts additional fatalities are implied although it is probable that the low loss of life was real, and is attributable to the second foreshock that evidently drew people into the safety of the open air fewer than 15 minutes before the mainshock.

In contrast to the anonymous accounts in Indian newspapers, Archibald Campbell, an assistant surgeon to the British Residency in Kathmandu, in three contributions to the Asiatic Society of Bengal (Prinsep, August 1833; Campbell, November and December 1833) compiled unofficial numerical damage and casualty data from the Kathmandu Valley, from travelers returning along the Kodari route from southern Tibet (passing through Dulka, the northern Nepalese border district of Dolakha), and from the official delegation returning from Lhasa through southern Tibet. He also mentions damage to forts along the southern trading route from India (Hetauda, Chisapani). Although he also obtained information from western and east-central Nepal his sources of information from these regions are not cited and data from these regions may be less reliable. Thus, unlike data from N and NE Nepal which mention specific towns and villages, his information from western Nepal refers to regional administrative areas. “At Gorkha (75km E of Kathmandu), only two houses were destroyed; at Palpa (180 km E of Kathmandu) none; and at Doti (540 km E of Kathmandu in easternmost Nepal), on the borders of Kumaon, the shock was felt, but by no means severely”. Although this westward decay in intensity over several hundred km is unsupported by details it is consistent with the decay in intensity documented independently in northern India. For example, Baird-Smith (1843 p.1051) reports “it was scarcely felt at all” at Lohughat (=Lohu Ghar, 29°23’N, 80°05’E, 18 km from the E. Nepal border). Campbell (1833) also emphasizes that the shock was not felt in Lhasa nor Xigatse, contrary to Mallet (1855, 238-240) and Dunn et al. (1939 p.116), who repeat early speculation by Prinsep (1833). In a following issue Prinsep, as editor of the Asiatic Society journal, corrected his erroneous guess that Lhasa was close to the epicenter (Campbell , 1833) but not until it had been repeated in Calcutta newspapers (e.g. India Gazette, Oct. 6 1833).

In general terms, information concerning the 1833 earthquake in northern India is restricted to British trading posts, and information from Nepal and Tibet appears to be restricted to the Kathmandu Valley and to villages on merchant routes from Kathmandu to Tibet and to India (Figure 3).

Fig. 2 Delay between mailing a letter and its arrival in Calcutta assuming publication occurs the day following its arrival. The three vertical lines indicate possible locations for Mullye. A=Murliganj, B=Manjhi and C=Malihabad or Malloon. The Vedic district of Malla lies between B and C. Distances are geodesic. The most remote point on the graph from Calcutta is Delhi.

Location names

Place names in India are readily matched with current locations for all but three 1833 listed localities: Tirhoot, Mullye (Mallai of Prinsep, 1833) and Baldeah. The first of these corresponds to the district of Tirhoot south of Nepal named after the town of that name now Muzaffarpur ( District Map of India, 1840, 64 mi=1”), however, it is not clear that the three damage reports from Tirhoot came from Muzaffarpur or from villages between Muzaffarpur and the Nepal border (Motihari, Sitamarhi, Darbhanga etc.). Despite the absence of specific locations, the high intensity shaking reported from Tirhoot unquestionably identifies a region in the southern Terai of Nepal or northern plains of India.

Mullye has eluded identification with either a district or town, and in this case the absence of a precise match is regrettable because from this region comes one of the two comparisons of the 1833 event with the 1803 Kumaun earthquake (the other account from a resident in Calcutta recalling the occurrence of a severe earthquake during the siege of Aligarh (27°55’N, 78°10’E) soon after the start of the second Anglo/Maratta war). The approximate location of Mullye is revealed by the fact that Calcutta news reports from Mullye took 9-11 days between mailing and publication. An additional clue is that a summary of the affects of the event south of 26°N from a correspondent in Agra omits mention of Mullye suggesting that it lies to the north of 26°N. Robert Mallet, who in 1851 discusses the 1833 earthquake as the 1834 Nepal earthquake (Mallet, 1852 p.313) places “Mallai?” on a travel time map of India (ibid. Plate XVII) at approximately 22.9°N, 91.2°E (Maijdi), yet the inferred intensities are too high, and the reported sun-clock times of the mainshock too early, for this location. He corrects the date to 1833 in his report of 1854 (Mallet, 1855, p.238). A graph of news travel times to Calcutta (Figure 2) shows that an eastward location for Mullye could correspond to the Vedic district of Malla, approximately between Gorackpur and Chapra, but this term is unlikely to have been in common use in the 19th century. Of towns and trading centers in the region, Mullye is considered a possible abbreviation for Malihabad or Malloon (Walker, 1833) near Lucknow, Manjhi (25°40’N, 84°30’E) near Chapra, or Murliganj (25°42’N, 86°59’E) near Purnia. Malihabad is evidently too far from Calcutta for normal 9 day news delivery, requiring travel times exceeding 120 km/day. Although the remaining suggested locations are at the correct range for news delays of 9-11 days, the phonetic resemblances are poor. In the absence of certain identification the three intensity V reports from Mullye are omitted from the isoseismal map.

It has not been possible to identify the precise location of Baldeah on the Kamla River where a flash flood followed the breaching of a temporary-lake dammed by an earthquake-triggered landslide. From the description (appendix) it is likely that this was near the Nepal border at longitude 86°08’.

Place names with phonetic similarity and geographic proximity to indicated regions have been identified for most of the villages in the Kathmandu Valley, and where these are questionable they have been identified in Table 1. Although relatively few locations outside the Kathmandu Valley are mentioned by Campbell, they are important in identifying shaking intensity to the north and east.

Campbell’s latitude for Tingri (28°N) is approximately ≈60 km too far south suggesting that he did not have good maps of southern Tibet. Rennell’s map of 1782 showing Tingri at 29.3N, was known to be in error to the British Residency in Kathmandu in 1801 but new maps prepared by Crawford in 1803 of the route to Dugurcheh (Xigatze) were very incomplete (Phillimore, 1950, p.71). Rennels map is of interest because it shows Nesty, Dunna, Kansa, Chuska, Kut or Kutí, and Mescingzhung in sequence northward toward Tingri, over the mountain pass between Nepal and Tibet near the present border village of Kodari. The spacing of villages suggests that each interval represents a day’s march. Nesty appears to correspond to Nisti, and the old fort at Dugunna Garhi may correspond to Dunna. Kansa is the village of Dram (Khasi in Tibet), 5 km NE and 400 m above Kodari corresponding to Kassa mentionedby Campbell (Nov. 1833) without explicit damage reports. Chuska of Rennell’s map may correspond to Choksum (28.07E,86.00N) or Kan Sing Chok (Campbell 1833). The location of Kuti mentioned several times by Campbell is the Tibetan name for Nyalam (D. Bresheares, personal communication 1994). At least two of the five villages mentioned as on the “Bhote or eastern pass to Tibet” (Nan Sing Chock, Kuti, Kassa, Mundun Pahar and Listigoan) appear to correspond to villages in a 20 km NS region of intense shaking that includes the present border village of Kodari, and it appears possible that information from a region within a quarter of a degree of 27°N, 28°E was conveyed by travelers returning from Tingri to Kathmandu along this trading route.

Fig. 3 Villages between Kathmandu and Tingri through the Kuti Pass (Rennell, 1782). Figure on right shows the true coordinate locations for Kathmandu, Tingri, Gyirong, inferred locations for Kassa, Sipa, Kuti, and merchant routes to India and Tibet.

Rennel’s map also shows the village of Sipa (Campbell’s village of Shipa, 32 km east of Kathmandu) on the Koƒs (Kosi) River between Sanku (Sankhu, 27°45’, 85°28’) and Ciopra (?Chautara). Sipa probably corresponds to Sipata or Sipatinghare (27°45’, 85°39’) on the Indrawatty tributary of the Sun Kosi which lies on the trekking trail to Kodari north of the current Kathmandu-Kodari road. The village of Nagarkot (meaning Snake Hill) on Rennel’s map between Ciopra and Nesty does not correspond to the geodetic observatory of that name on the eastern hills overlooking the Kathmandu valley.

The location of places in Campbell’s Dec. 1833 account are of particular importance in tracing the felt area to the north and east of Kathmandu. Digarchi is the present Shigatse or Xigatze (29°17’, 88°54’) and Tingri (28°30’, 86°30’) retains its old name. However, from Tingri Campbell indicates that a 8-10 day march due-west brought the Nepal/Beijing returning delegation to the village of Kirung. Kyirong or Gyirong (28°33’, 85°16’, Shiron on Rennel’s map) is 180 km W. along the road from Tingri, and although the pass south of Kyirong enters Nepal by the Bhote Kosi, which means “Tibet sacred river” it is only one of several rivers of that name crossing the border, including one through the Kuti pass into Kodari. It is for this reason that Campbell may have been mislead into entering in parenthesis that the delegation came through “the eastern pass of the valley into Bhote”. The Kyirong route they followed would more appropriately be called the northern pass to Tibet.

Isoseismal map for 1833

Descriptions from the original reports were assigned values on the Modified Mercalli Intensity scale (Tables 1 and 2 and Appendix). Where several reports from the same village exist, the intensity data in most cases agree adding confidence to the observations Moreover, although the intensity data for the 1833 event are somewhat uneven in coverage they yield relatively smooth spatial variations in intensity (Figure 4). The intensity data may be interpreted as smooth curves that approximately encircle the appropriate intensity data, or more complex figures that adhere rigorously to mapped intensities. Figure 4 shows a series of smooth ellipses that approximate the observations. The resulting isoseismals are prejudiced by a preference for simplistic geometry, yet although a number of alternative curves could be constructed they would have the unavoidable characteristic of indicating maximum intensities in east-central Nepal, close to the 1934 epicentral region. A more realistic set of isoseismal curves is considered unwarranted given the distribution of intensity data. As an example of the complexity denied by adopting elliptic isoseismals, the intensity observations from the 1833 earthquake are superimposed on a map of the observed and inferred isoseismals for the 1934 Bihar earthquake. Few of the 1833 intensity data are found outside the corresponding 1934 isoseismals suggesting that similar complexity in local isoseismal geometry may have prevailed in the two events, enforced by basin response characteristics.

An approximate estimate for the felt area of the 1833 earthquake (dashed on Figure 4) is found in Oldham’s memoir on the 1897 earthquake (Oldham 1899). Oldham does not indicate his sources although it is likely that the data come from his father’s chronology of Indian earthquakes (Oldham, 1883), and that the contour shown is meant to represent the area within which the earthquake was perceptible (i.e. Mercalli intensity II). This 0.21x10⁶ km² felt area includes Chittagong to the southeast (editorial footnote by Prinsep in Campbell, December 1833, “Mr. Walters informs me that it was also felt in Chittagong”), but omits information from Delhi and other western points, and from Jabalpur to the SW. Using a smooth ellipse to embrace these western points increases the felt area to approximately 1.2x10⁶ km². Uncertainty in this estimate is caused largely by the sparcity of data to the SW, E and N. Although the elliptical aspect ratio of the intensity ≥VI isoseismals is favored by the specific absence of felt shaking in Lhasa and Xigatse and by the shape of intensity VII isoseismals, the absence of reports from central India in theory permits the felt radius to increase substantially to the SW. Thus the estimated areas for intensity<V are assumed to err on the low side and estimates for earthquake magnitude presented below are given both for elliptical and circular isoseismals. Intensity I-III isoseismals are omitted from Figure 4 as no data are available. The largest ellipse drawn for the 1833 earthquake in Figure 4 corresponds to Mercalli Intensity IV which is presumed to be significantly smaller than the area of felt perceptibility. Reports from locations specifying that the earthquake was not felt are absent except to the NE, and from remote locations such as Bombay and Madras.

Uncertainties in isoseismal area in Table 1 correspond to maximum and minimum estimates for areas contained within the smooth symmetrical contours selected to fit the observations in Figure 4. Although these are apparently uncertain to ±1% based on the mapped isoseimal reports, this figure does not account for errors in estimating the Mercalli Intensity from the eye witness accounts in the Appendix. Variations in building toughness and subjective interpretation of damage may result in errors perhaps an order of magnitude larger. Moreover, from the 1934 (Dunn et al. 1939) and 1988 (Sinha, 1993) Bihar/Nepal events it is clear that that isoseismal contours in the region are typically more complex.

The congruence between 1833 and 1934 intensity data is interesting in that in the few locations that isoseismic intensities disagree they differ by not more than one intensity unit. The 1934 isoseismals shown in Figure 4 are somewhat uncertain within Nepal since they are formed by merging the isoseismals shown in Dunn et al. (1939) with intensity data discussed by Pandey and Molnar (1988). Although the resulting extension of the intensity >VIII data northward can be justified using the authority of Rana (1934) the position and dimensions of the intensity IX region is conjectural. The centroid of intensity VIII area for the 1833 earthquake is apparently displaced approximately 1 degree to the west of the equivalent region for the 1934 earthquake, consistent with Khattri’s 1987 location. This observation is sustained, however, by three observations only: at Goruckpur, Gorkha and at Chapra (Chuprah) on the Ganges. Examination of the first two of these reports shows that intensity VIII data are but weakly supported by the reports. Thus, were the buildings damaged of poor construction, a lower intensity could be assigned to these villages. The 1833 Chapra account describes slumping (“a chasm of considerable depth formed”) which may indicate a localized region of high intensity shaking near the Ganges similar to the narrow intensity IX region in 1934. The inclusion of these westernmost VIII intensities in an elliptical fit to this isoseismal results in an 1833 intensity VIII area similar to the inferred 1934 intensity VIII isoseismal. However, a much smaller region for 1833 intensity VIII is admitted by the data if the same pattern of localized severe shaking occurred as in 1934. For example, the sparcity of intensity VIII data in 1833 between Dhankuta and Dharbanga admits the possibility that the area of intensity VIII shaking could be as low as 60,000 km² or as high as 100000 km2, the latter being the inferred area of intensity VIII shaking in 1934.

Three regions of severe damage were reported in the 1833 and 1934 earthquakes, an observation that led Dunn et al. (1939) to note a similarity between the two events: the Kathmandu Valley, a high intensity region near the Ganges including Monghyr, and a region north of Muzaffarpur. Data near the slump belt of 1934, a region of catastrophic lateral spreading, is sparse and although high intensities are recorded near Tirhoot and Purnea (VIII-IX) no liquefaction features are mentioned here in 1833. The area consists of flat lying sediments and ox-bow lakes across which numerous rivers meander. Further south, at Chapra and Monghyr, accounts indicate localized liquefaction near the Ganges, and minor ground damage is reported to the north, near Bhagmati in the Kathmandu Valley. Shaking intensities in the Kathmandu valley were similar for both earthquakes with highest intensities near Patan. Baird-Smith (1843) noted that damaging shaking at Monghyr frequently accompanies large earthquakes, an observation repeated both by Dunn et al. 1939, and by Pandey and Molnar 1988. A favored explanation for localized high intensities in this region is that surface waves are amplified in the water-saturated sediments as they approach the southward shelving bedrock surface south of the Gangetic Plain (ibid, 1988). Similarly, the lake deposits of the Kathmandu valley can be assumed responsible for localized high intensities and rapid variations in intensity in this region as were observed most recently in 1988 (Dikshit and Koirala, 1989).

Magnitude of the 1833 Earthquake

Various empirical relations have been developed to relate felt areas to magnitude (e.g. Toppozada, 1975; Hanks and Johnston 1993; Johnston, 1994a). A physical formulation to account for the observed relation between intensity and moment magnitude has been proposed by Frankel (1994) the coefficients of which have been fit to the global data base by Johnston (1994b). I shall refer to this as the F94 model. Moment magnitude, Mo, of an earthquake in the F94 model is related to the area enclosed within a specified isoseismal intensity contour, S, with an expression of the form

log Mo = a + blogS +c√S (1)

where the constants a, b and c are determined empirically for each isoseismal area. Intensity magnitudes are shown in Table 3 using coefficients derived from worldwide data that include 6 Indian events (Johnston, 1994b). A mean magnitude determined for the 1833 earthquake in this way is 7.5±0.3, close to the M=7.6 adopted by Khattri and Tyagi (1983), however, a monotonic decrease in estimated magnitude with intensity is observed, presumably indicating systematic differences between the global coefficients and those appropriate for the 1833 event. Data for several Indian earthquakes are presented in Figure 6 to illustrate these possible biases. Notwithstanding the ensuing discussion, a cursory inspection of Figure 6 indicates that the 1833 earthquake must be greater than M=6.6 and less than M=8.1, the inferred magnitudes of the Bihar 1934 and Udaypur 1988 earthquakes.

______________________________________________________________________________

Table 3 Isoseismal Areas and estimated Moment Magnitudes for the 1833 earthquake.

Constants a, b, and c are from Johnston (1994) for global data least-squares fit to equation (1). Moment magnitudes for elliptical areas (F94e) are from Figure 4, and those for circular areas (F94c) for diameters equal to the long axes of the ellipses in Fig. 4 where unconstrained

by data to the SE. Areas are expressed as log₁₀(area km²) and Mo in log₁₀(dyne cm).

______________________________________________________________________________

Intensity felt IV V VI VII VIII mean Mo

a 19.62 18.36 20.44 19.51 22.96 24.3

b 0.5 0.903 0.607 1.307 0.00307 0.00655

c 0.00163 0.00206 0.00312 0.00079 0.00307 0.00655

elliptical area (E) 6.16±.08 5.87±.06 5.55±.02 5.35±.02 5.02±.04

log(Mo) F94E 26.4 26.7 27.2 27.8 28.0

Mo F94E 6.9±0.2 7.1±0.1 7.46±.02 7.82±.03 7.9±0.3 7.5±.3

circular area(C) 6.42 6.16 5.86

Mo F94C 7.6 7.9 7.9 7.8±.2

____________________________________________________________________________

Frankel (1994) assumes that attenuation is radially symmetric resulting in circular isoseismals. These are not appropriate for the Himalayan plate boundary. Yet it is not clear whether the observed isoseismal ellipticity is an artifact of reporting, an east-west amplification effect, or a north-south attenuation effect. The elliptical isoseismals inferred in Figure 4 have an approximate 2:1 aspect ratio with their long axes parallel to the Himalaya, but as mentioned previously this is possibly biased by the absence of reports to the north and to the SE. By assuming that the lower intensity isoseismals approximate circles with their radii equal to the semi-major axis of the ellipses shown in Figure 4, circular intensity IV, V and VI isoseismals can be invoked that yield intensity magnitudes of 7.6, 7.9 and 7.9 respectively, raising the mean magnitude intensity to M=7.8. Although the observations permit this interpretation to the SE, the “not felt” data from Lhasa and Xigatse (Campbell, 1833) are inconsistent with circular isoseismals. Moreover, the elliptical aspect of Himalayan isoseismals appears to be a common feature of Himalayan earthquakes. Intensity VI to VIII isoseismals for the M=6.6 Udaypur 20 August 1988 earthquake exhibit an aspect ratio of approximately 4:1 (Pandey and Nicolas, 1989; Dikshit and Koirala, 1989). Isoseismals VI-VIII for the 1991 Uttarkashi earthquake are elongated 2:1 along arc (Thakur and Kumar, 1991) and intensity >VI and VII isoseismals for other Himalayan events show elongation parallel to the arc (e.g. Kangra VII 1905, and Bihar 1934). Thus although the absence of radial symmetry for intensities >VIII could be ascribed to source geometry, it is surely necessary to invoke transmission effects as the reason for the elliptical isoseismal areas VI-VII found in the southern Himalaya and Gangetic Plain. For example, it is likely that the sediments overlying the stable continental crust in the Gangetic Plain act as a wave guide for surface waves traveling east-west with periods of 1-3 s (Banerji, 1953). Similarly, the dipping interface between the Indian plate and the overlying sediments may result in less interference along strike than updip and downdip, favoring along-arc perceptibility. For the lower intensities that are felt south of the Gangetic Plain the intensity distributions for the four great earthquakes between 1897 and 1950 are approximately circular, consistent with transmission within uniform crust.

Figure 5 Moment magnitudes inferred from Johnston’s F94 coefficients for several earthquakes in Nepal and India. The shaded area represents the range of magnitudes inferred for the 1833 earthquake using elliptical isoseismals (Fig. 4). The overlying fenced region assumes circular isoseismals (see text). Assam 1897 data from Oldham (1899). Bihar 1934 from Dunn et al. (1939) and Pandey and Molnar (1990). Udaypur; Dikshit and Koirala, (1989); Sinha (1993). Events in tilted script from Johnston (1994).

Epicenter, Aftershocks and Rupture areas in 1833 and 1934

The location of the mainshock cannot be determined unequivocally from the data presented here. This is largely because most of eastern Nepal and southern Tibet are unrepresented in the felt report data, preventing incontestable closure of isoseismals greater than VIII in these directions. A tantalizing intelligence attributed to “vulgar report” by Campbell (Nov. 1833), is that for 5 days before the earthquake “noises similar to the firing of cannons were heard as if underground: and in this neighborhood” (Kan Sing Choke, possibly Choksum at 86E, 28.07N), “the high road to Lhasa is said in many places to be blocked up by the fallen earth from the mountains”. If these are interpreted as foreshocks close to the epicenter of the mainshock, the epicenter would be placed unreasonably close to a rapid decay in felt intensities eastward. That is, intensities at Tingri 100 km to the NE are estimated at IV, whereas at Kyirong and Kathmandu, similar distances from Choksum to the NW and SW, intensities exceed intensity VIII. Thus if the mainshock, or one of the foreshocks, occurred near Choksum, this may have been the eastward limit of subsurface rupture. The few aftershocks that were reported in 1833 do not provide any clear idea of the possible rupture zone of the earthquake southward or westward although their widespread manifestation favors a shallow thrust more than than a deep focus event such as the 1988 Udaypur earthquake. Continuous ground motion in the days following the 1833 earthquake was claimed at Monghyr, Tirhoot and Kathmandu, the three regions of highest intensity shaking. The concentration of felt aftershocks within these areas may not indicate the proximity of aftershocks to these locations, but instead may indicate either that the frequency of shaking in these locations was optimal for people to sense (e.g. 2-4 Hz, Frankel 1994), or that anxiety about further destruction heightened their awareness to aftershocks. On 4 October 1833 a strong aftershock was felt simultaneously at Kathmandu, Allahadad, Berhampore, Malda, Purneah and Bhagalpore, and another on 18 October at Goruckpore, Kathmandu and Allahabad. A surprising number of aftershocks were felt in the slump belt region in 1934 and the largest instrumentally located aftershock in 1934 is also found in this region (Dunn et al. 1939).

Seeber and Armbruster (1981), and Chander (1989) propose that the 1934 earthquake ruptured a region forward of the main frontal thrusts of the Himalaya, at least as far south as the slump belt. High intensity shaking was also observed near the Greater Himalaya and central Nepal near the instrumentally determined epicenter approximately 100 km north of the slump belt (Pandey and Molnar, 1990). If it is assumed that the 1934 earthquake was Mo=8.1 its rupture area could easily extend embrace these two regions and extend an equivalent distance, or greater, along strike. However, the 1833 earthquake with a moment magnitude of 7.6-7.8 is unlikely to have ruptured an area more than 70 km on a side. The absence of catastrophic damage at the forts of Chisapani and Mukwanpur argues against a rupture zone centered at or near latitude 27°N, and it seems unlikely that rupture further south (beneath the plains) could generate >30% destruction in towns in southern Tibet. Thus the rupture zone of the 1833 earthquake appears to have been centered in northern Nepal or southern Tibet, and unless it was a thrust of unusual aspect ratio it would not extend beneath northern India. Thus the inferred intensity VIII and IX shaking in the Gangetic plain may have resulted from basin resonance, and not from the underthrusting invoked to have occurred during the 1934 event.

Following this line of reasoning, if a M=7.7 1833 earthquake near the Greater Himalaya could generate intense localized shaking in the Gangetic Plain, so too could the Bihar 1934 earthquake. Could the slump belt in 1934 be a product of basin resonance and prolonged shaking thus not requiring slip to extend beneath the Gangetic Plain? Catastrophic lateral spreading occurred in the Alaska 1964 earthquake and was the result of vigorous and prolonged shaking of saturated sediments. It is not difficult to imagine that catastrophic lateral spreading could be confined to a region of especially intense shaking in 1934. The removal of the requirement for slip terminating near the slump-belt would remove the obvious difficulty that there is no evidence for a topographic bulge above the required southern edge of the inferred rupture. This was not revealed by leveling, nor is there any topographic evidence to suggest that the ruptures of previous shallow thrust earthquakes have terminated near the slump belt.

If it is assumed that the 1833 and 1934 earthquakes were both shallow thrusts, a limited number of geometries are possible if we additionally require that the two ruptures do not overlap. However, this requirement is not essential because the rate of development of available slip from Indian/Tibetan plate convergence is 1.5-2 m per century (Molnar, 1990), and slip released in the 1833 event could have been effectively renewed by 1934. The simplest relation for the two events is that a region north of Kathmandu ruptured in 1833 and that a region roughly four times the area ruptured in 1934 immediately to the east. However, alternative and not necessarily more complex rupture geometries can be envisaged that would also be consistent with the available data.

Isoseismal areas and Great Himalayan Earthquakes

Were the worldwide coefficients determined by Johnston (1994) for equation (1) appropriate to relate the felt areas of Indian earthquakes to their moment magnitudes, each event in Figure 5 would be represented by an horizontal line. However, intensity magnitudes determined for the four largest events show significant variance, indicating that the data for these events are unreliable, or that the Eurasian/Indian plate boundary requires adjusted coefficients in equation (1). Unfortunately the intensity data for the four great earthquakes are both confusing and incomplete. They are confused partly because intensity scales developed for the industrial nations are not easily applied to the type of construction and destruction in each area, and partly because those responsible for the compilation of intensity data were unable to visit much of the afflicted areas to examine inconsistences in written reports and hence apply uniform critera to the observed damage. They are incomplete because isoseismal coverage is typically only for one half of the shaken region. The data in Figure 5 have area uncertainties of up to 50%.

For all four great earthquakes the maximum “felt area” isoseismal reported in the standard works on these earthquakes (Oldham, 1899; Middlemiss, 1905; Dunn et al. 1935; Tandon, 1953) yield moment magnitudes 1-2 magnitude units lower than the magnitude derived from the intensity IV isoseismal. Given that the intensity IV to VI isoseismals yield moment magnitudes consistent with those derived from radiated seismic energy, a possible conclusion is that the “felt area” isoseismal has been significantly underestimated for these events. Surface wave dispersion results in progressively lower frequencies dominating the accelerations at increasing distances from the epicenter and distant accelerations are presumably shifted to frequencies that are not optimum for human perception. Distant “felt” reports thus become increasingly “instrument”-based: long period oscillations in reservoirs, movements of chandeliers in churches, and slow movements of doors.

The 1897 isoseismals documented by Oldham 1899 consist of precisely delinated regions of damage to masonry structures, and approximate areas assigned grouped Rossi-Forrel Intensities. The low intensity isoseismal areas are completed by extending Oldham’s elliptic isoseismals to the north, however, the resulting areas may be uncertain by 50%. Isoseismal areas listed by Tandon (1953), Poddar (1953) and Ray (1953) for the Assam 1950 earthquake yield systematically low moment magnitudes unless these are doubled to include the areas not visited to the north and east of the epicenter. This extrapolation is undertaken “with the risk of great departure from the truth” (Ray,1953), a truth already uncertain for the same reasons that the 1897 isoseismals are unreliable. Estimates of lower and upper bounds for these isoseimals results in more than a ±0.5 variation in assessed moment magnitude. Both Assam earthquakes yield mean moment magnitudes between 8 and 9.

The 1934 Mercalli intensity data reported by Dunn et al. (1939) yield reasonably consistent moment magnitude estimates although the values are somewhat higher than those derived by Johnston (1994). This is presumably because of uncertainties in closing intensity IV-VI isoseismals in southern Tibet. The intensity VIII area adjusted in Figure 4 to merge felt data from north and central Nepal (Pandey and Molnar, 1990) with data from Dunn et al. (1939) yield a good approximation to the observed moment magnitude.

The Kangra isoseismal data as reported by Middlemiss (1910) are unique among the four great Himalayan events in that they cannot easily be reconciled with the Johnston F94 coefficients. In particular, the mapped intensity VIII data are confined to two small regions, yielding an absurdly low estimate for moment magnitude (M≈7) compared with hitherto accepted magnitudes for the event (M=8.4). This anomalously small area is evident in Figure 1. Yet the estimated intensity V isoseismal area for this event is consistent with a M≥8 earthquake. Even were the two mapped intensity VIII areas to connect, which Middlemiss (1910) is adamant they do not, or the Kangra intensity VIII isoseismal to bend southward, as suggested by Molnar (1987), the increased area would bring the moment magnitude to no more than M=7.5. A localised intensity VIII area associated with a large earthquake would normally be interpreted as symptomatic of a shallow earthquake, or an earthquake in a region with abnormally high attenuation. Yet there no surface rupture was reported, and the widespread felt area indicates that the magnitude of the earthquake is typical of other great Himalayan events. The inversion of leveling data from the easternmost area is consistent with a slip on a shallow thrust of 7.5 m (Galalaut et al., 1994), again consistent with the occurrence of a great earthquake. A possible explanation for a small high intensity region and a large felt area is that the Kangra 1905 event was a slow earthquake. Slip on a shallow dipping buried thrust may have occurred at rates lower than the 1897, 1934 and 1950 earthquakes, insufficient to generate a widespread region of high accelerations but sufficient to stimulate long period waves resulting in a large felt area. In this interpretation, the high intensities observed near Kangra and Dehra Dun may have been associated with secondary shallow faulting in response to underlying slip.

Conclusions

A 1 million km² of northern India was shaken by a significant earthquake in 1833 that appears to have been centered in north-central Nepal. Although intensity X shaking occurred locally and more than 4000 buildings are reported destroyed, the loss of life (≈500) was small largely due to the occurrence of two large foreshocks hours and minutes respectively prior to the mainshock. A possible location for the epicenter of the 1833 mainshock is approximately 50 km north, or north east of Kathmandu, although the limited number of observations in east central Nepal and southern Tibet permit an epicenter to the east of Kathamandu, close the epicenter of the 1934 Bihar earthquake. The 1833 earthquake resulted in damage intermediate in severity to the Udaypur M=6.6 1988 and Bihar M=8.1 1934 earthquakes. A mean moment magnitude of Mo=7.7±0.2 is obtained by applying Johnston’s F94 1994 relation between isoseismal areas and moment magnitudes.

The inferred M<8 magnitude and consequent relatively small rupture area of the 1833 earthquake indicates that it contributed insignificantly to reducing potential slip in the Central Himalayan Seismic Gap between the Kangra 1905 and Bihar 1934 rupture zones. The slip associated with the event may have been 1-2 m, an amount consistent with a renewal time of 100 years. This and its proximity to the inferred rupture zone of the Bihar 1934 earthquake suggests that if the 1833 event occurred on a thrust fault it may have ruptured a region abuting or overlapping the 1934 rupture. However, although remarkable similarities between the 1934 and 1833 isoseismal areas exist, the mechanism of the 1833 event is unknown.

The application of the F94 Johnston (1994) algorithm to intermediate intensity isoseismal areas of four great earthquakes in the Himalaya yield moment magnitudes consistent with currently accepted values. Maps for isoseismal intensities less than V and greater than VII are either less reliable or incomplete for each of these earthquakes. Even when generous allowance is made for the northward extension of the perceptible felt area the intensity II area for the four events yield values that are too low by more than 1 magnitude unit, indicating that recalibration of the F94 algorithm for Indian earthquakes is desirable.

The Kangra 1905 earthquake is unique in that its intensity VIII felt area is anomalously small for a great earthquake. The seismically determined magnitude, the geodetically estimated slip, and the intensity IV-VI felt areas are consistent with an 8.5>Mo>8 event whereas the Intensity VIII area is appropriate for an Mo≈7 event. A possible explanation for this anomalously small area of intensity>VIII shaking is that the earthquake occurred with a substantial slow component, and that the observed high intensities were associated with secondary ruptures near Kangra and Dehra Dun. The absence of broad-band seismic data for this and other Himalayan earthquakes makes such a conclusion tentative. However, if slow earthquakes do occur in the Himalaya they would have the benefit of absorbing plate convergence with less high intensity shaking than normal high frequency earthquakes. Moreover, if other M≈7 earthquakes in the historic record represent the secondary seismic manifestations of slow great earthquakes the maturity of the Central Seismic Gap may not as advanced as currently believed.

Acknowledgements

I thank Arch Johnston for providing me with an advanced copy of his forthcoming paper, and Vinod Gaur for materials and insight on aspects of Himalayan seismicity. Librarians at the Norlin Library, Boulder, Colorado and the India Office Library, London, were of great assistance in locating historic materials related to the 1833 earthquake. The research was funded by the National Science Foundation.

References

Abe, K., 1994. Instrumental magnitudes of Historical Earthquakes, 1892 to 1898, Bull. Seism. Soc. Am. 84(2), 415-425.

Andrews, C. F., 1935. The Indian Earthquake, George Allen and Unwin, pp.135.

Anzidei, M., 1994, GPS surveys in eastern Nepal, Terra Nova, 6, 82-89.

Arya, A. S., 1992, Possible effects of a major earthquake in Himachal Pradesh, Seismology in India, Current Science, 62, 251-256.

Baird-Smith, R. 1843, Memoir on Indian Earthquakes, Pt. II, J. Asiatic Soc. Bengal, Vol. 12(2), New Series, 136, 1029-1056†. {{Note to printer-The symbol † is required}}

Baird-Smith, R., 1844, Memoir on Indian Earthquakes, Pt. III, J. Asiatic Soc. Bengal, Vol. 156, 964-983.

Banerji, 1953, The origin and the nature of disturbance produced by the Assam earthquake of Aug. 15 1950, and its aftershocks, 11-15, in A compilation of papers on the Assam Earthquake of August 15, 1950. ed. M. B. Ramachandra Rao, Publication No. 1 Central Board of Geophysics, Gov. of India, 1953.

Bapat, A, Kulkarni, R. C. and S. K. Guha, 1983, Catalog of Earthquakes in India and Neighborhood from historical period up to 1979, Ind. Soc. Earthq. Tech. Roorkee, pp. 211.

Beroza, G. C. and T. H. Jordan, 1990, A search for slow and silent earthquakes using free oscillations, J. Geophys. Res. 95, 2485-2510.

Bhat, M. R., 1985, (trans.) Brhat Samhita by Varahamihara, English Translation, Exhaustive notes and literary comments, Motilal Banarsidass, Delhi, Pt. 1.

Bilham, R., P. Bodin and M. Jackson, Entertaining a Great Earthquake in Western Nepal - Historic Inactivity and Geodetic tests for the development of strain. J. Nepal Geol. Soc. In the press 1995.

Calcutta Gazette, 1803, The Calcutta Gazette, , XL, 1019 Sept. 8 and 1020, Sept 15.

Campbell, A, 1833, Further particulars of the Earthquake in Nepal, J. Asiatic Soc. Bengal, 2, Misc. VI, 636-639, Dec. 1833

Campbell, A., 1833, Account of the Earthquake at Kathmandu, J. Asiatic Soc. Bengal, 2, Misc. II, 564-567, Nov. 1833.

Chander, R., 1988, Interpretation of observed ground level changes due to the 1905 Kangra earthquake, northwest Himalaya. Tectonophysics, 149, 289-298.

Chander, R., 1989, Southern limits of major earthquake ruptures along the Himalaya between longitudes 75° and 90° E., Tectonophysics, 170, 115-123.

Chandra, U., 1992, Seismotectonics of Himalaya, in Seismology in India, Current Science, Indian Acad. Sciences, Bangalore. 62, 40-71.

Chen, W. P. and P. Molnar, 1977, Seismic Moments of Major Earthquakes and the average rate of slip in Central Asia, J. Geophys. Res, 82, 2945-2969.

Dikshit, A.M. and A. Koirala, Report on the Intensity mapping of Udaypur Earthquake of 20 August 1988. HMG Ministry of Industry, Nepal , Dept. of Mines and Geology, Lainchaur, Kathmandu May 1989

Dunn, J. A., J. B. Auden, A. M. N. Gosh and S. C. Roy, 1939 (reprinted 1981), The Bihar-Nepal Earthquake of 1934, Mem. Geol. Soc. of India, Survey of India, Calcutta. 73, pp. 391.

Frankel, A., 1994, Implications of Felt Area-Magnitude relations for earthquake scaling and the average frequency of Perceptible Ground Motion. Bull. Seism. Soc. Amer. 84(2), 462-465.

Gahalaut, V. K. and R. Chander, (1992), A rupture model for the great earthquake of 1897, northeast India, Tectonophysics, 204, 163-174.

Gahalaut, V. K., P. K. Gupta, R. Chander, and V. K. Gaur, 1994, Minimum norm inversion of observed ground elevation change for slips on the causative fault during the 1905 Kangra Earthquake. Indian. Acad. Sciences. in the press.

Gaur, V. K., 1980, Earthquake risk to Tehri Dam, Rep. Natl. Geol. Res. Inst., Hyderabad, 22 pp.

Gaur, V. K., ed. 1993, Earthquake Hazard and Large Dams in the Himalaya, Indian Trust for Art and Cultural Heritage, New Delhi, pp. 152.

Gupta, G. D. ed. 1992, Himalayan Seismicity, Geol. Soc. of India, Bangalore, Memoir 23, pp. 334.

Gupta, H. K. Earthquake Prediction Related Studies in India, Seismology in India, Current Science, Indian Inst. Sci. Bangalore, 62, 257-264, 1992.

India Gazette, The India Gazette, Calcutta, III, 3 Sept.-26 Oct. 1833.

Iyengar, R. N., 1994, Earthquake History of South India, The Hindu, Jan. 23.

Jackson, M., and R. Bilham, 1994a, 1991-1992 GPS Measurements across the Himalaya, Geophys. Res. Lett., 21, 1169-1172.

Jackson, M., and R. Bilham, 1994b, Constraints on Himalayan Deformation Inferred from Vertical Velocity Fields in Nepal and Tibet, J. Geophys. Res., 99, 13897-13919.

Johnston, A. C. Electric Power Institute, Palo Alto.Johnston, A. C. Moment Magnitude Assessment of Stable Continental Earthquakes, part 2: Historical Seismicity, Geophys. Journal International. in press 1994.

Kanamori, H and D. L Anderson, 1975. Theoretical Basis of some Empirical Relations in Seismology, Bull. Seism. Soc. Amer. 65(5), 1073-1095.

Khattri, K. N, and A. K. Tyagi, 1983, Seismicity Patterns in the Himalayan Plate Boundary and Identification of Areas of High Seismic Potential, Tectonophysics, 96, 281-297.

Khattri, K. N., 1987, Great earthquakes, seismicity gaps and potential for earthquake disaster along the Himalaya Plate boundary, Tectonophysics, 138, 79-92.

Khattri, K. N., 1992, Seismic Hazard in Indian Region, in Seismology in India, Current Science, Indian Inst. Sci., Bangalore, 62, 109-116.

Mallet, R., 1852, Report of the Twenty-first meeting of the British Association for the Advancement of Science, Ipswich, 1851, Second Report on the Facts of Earthquake Phenomena, 272-320

Mallet, R., 1855, Report of the Twenty-fourth meeting of the British Association for the Advancement of Science, Liverpool, 1854, Catalogue of Recorded Earthquakes from 1606 B.C. to A. D. 1850 (continued from Report for 1853), 2-326.

Middlemiss, C. S., 1910, The Kangra Earthquake of 4 April 1905. Mem. Geol. Soc. of India, 38, pp. 409, Geol. Surv. India, Calcutta.

Molnar, P., 1984, Structure and tectonics of the Himalaya: Constraints and implications of geophysical data, Ann. Rev. of Earth and Planetary Sci., 12, 489-518.

Molnar, P., 1987, The Distribution of Intensity Associated with the 1905 Kangra Earthquake and Bounds on the Extent of the Rupture Zone. J. Geol. Soc. India, 29, 221-229.

Molnar, P., 1987, The Distribution of Intensity Associated with the Great 1897 Assam Earthquake anmd Bounds on the Extent of the Rupture Zone. J. Geol. Soc. India, 30, 13-27.

Molnar, P., 1990, A review of the Seismicity and the Rates of active underthrusting and te deformatation at the Himalaya, J. Himalayan Geol., 1, 131-154.

Molnar. P and M. R. Pandey, 1989, Rupture zones of great earthquakes in the Himalayan Region. Proc. Ind. Acad. Sci., (Earth and Plan. Sci.) 98, 61-70.

Oldham, R. D., 1899 (1981 reprint), Report on the Great Earthquake of 12 June 1897, Mem. Geol. Soc. of India, 29, pp. 379. Geol. Surv. India, Calcutta.

Oldham, T., 1883, Catalog of Indian earthquakes, Mem. Geol. Surv. India, 19, 163-215, Geol. Surv. India, Calcutta.

Pandey M. R. and M. Nicolas, 1989, Dept. of Mines and Geology, HMG Nepal, Report 2 March 1989, The Aftershocks sequence of the Indian-Nepalese earthquake of August 20, 1988,

Pandey, M. R. and P. Molnar, 1988, The distribution of Intensity of the Bihar Nepal earthquake of 15 January 1934 and bounds on the extent of the rupture, J. Nepal Geol. Soc, 5, 22-44.

Phillimore, R. H., Historical Records of the Survey of India, 2, 1950