- © Seismological Society of America
We present macroseismic analyses of three historical earthquakes that occurred in 1750, 1838, and 1904 in the Bakar epicentral area (Rijeka region, Croatia). Using various historical sources, we were able to compile intensity maps and assess macroseismic parameters for each of these events. The estimated epicenters lie close to known active fault systems, and the epicentral intensities are lower than those listed in catalogs. We have also identified a number of strong foreshocks and aftershocks which are currently missing from all catalogs. The three events analyzed, together with the Klana earthquake of 1870, are the largest known events to have occurred in these tectonically active parts of the northwest Dinarides. As such, they are important in characterizing seismic hazard in the vicinity of the main Croatian port, the city of Rijeka.
Tables of localities, intensities, and sources
The Croatian city of Bakar lies at the southeasternmost part of the Rijeka epicentral region, which extends to the north to Ilirska Bistrica in Slovenia (Fig. 1a). It is a seismically active part of the northwest External Dinarides where strong earthquakes are known to have occurred in the past. Seismicity along the northeast Adriatic coast (Fig 1b) is shallow (Herak and Herak, 1990) and is generally assumed to be caused by the ongoing underthrusting of the Adriatic microplate (Adria) beneath the Dinarides along the northeast coast of the Adriatic Sea and by the Adria’s collision with the Alps in the north. Adria rotates counterclockwise at a rate of about 0.25°/My around a pole in northern Italy, driven by the push of Africa from the south (see e.g., Anderson and Jackson, 1987; Battaglia et al., 2004; Serpelloni et al., 2005; D’Agostino et al., 2008; Weber et al., 2010). More about the current understanding of the rather complicated interaction of tectonic units here and their evolution over time may be found in, for example, Handy et al. (2015). The Adria–Europe convergence resulted in a thrust front along the entire northeast Adriatic coastline, although the activity and the amount of shortening are much larger in central and southern Dinarides than it is in their northern part. The External Dinarides are derived from the Adriatic Carbonate Platform, and hence limestone, dolomites, and breccia dominate in its northwest part. The whole area is intersected with the karst fields of different sizes, often very narrow and elongated along the Dinaric strike, that are filled mostly with Quaternary alluvial and proluvial deposits. These may cause amplification of seismic waves, which results in higher macroseismic intensities in the karst fields. The islands are mostly built of Cretaceous carbonates. The seismicity and seismotectonic activity here was studied by Aljinović et al. (1984), Blašković (1999), Prelogović (1989), Prelogović et al. (1980, 1981, 1995, 1998), Živčić and Allegretti (1983), and Tomljenović et al. (2010). Seismicity overviews in the larger context of Croatian seismicity can be found in Milošević (1980), Herak et al. (1991), Markušić et al. (1990, 1993, 1998), and Ivančić et al. (2002, 2006).
Insight into historical seismicity provides lower bounds on the deterministic seismic hazard of an area, which is most important in territories with low instrumentally recorded seismicity. Historical earthquakes from the northwest Dinarides have so far been rather poorly studied. Work on the compilation of earthquake catalogs with epicenters in Croatia and the surrounding areas began under the United Nations Development Programme (UNDP)/United Nations Educational, Scientific and Cultural Organization (UNESCO) project on seismicity in the Balkan region for the period 1901–1970 (part I) and before 1901 (part II) (Shebalin et al., 1974), accompanied by part III (Shebalin, 1974), which presented 493 isoseismal maps of strong earthquakes (only one from the epicentral area studied here). These publications were adopted as nuclei for the current Croatian earthquake catalog (CEC, updated version first described in Herak et al., 1996; see also Data and Resources) and the Croatian macroseismic database (updated version described in Sović, 1999; see also Data and Resources), which are being constantly revised (including relocation of all instrumentally recorded events since 1906 when the Zagreb station was founded) and updated with new entries on a yearly basis. However, no such systematic effort has been carried out concerning the preinstrumental data, that is, the isoseismal maps of important events from the past. Notable exceptions regarding individual events are the works on the (fake) earthquakes near Zagreb in 1502 (Cecić et al., 1998) and Rijeka in 1802 (Albini et al., 1994).
The Rijeka epicentral region (Fig. 1) is seismically active, which is reflected in relatively high hazard estimated there (exceeding 0.25g on bedrock for the return period of 475 years, according to the Croatian seismic‐hazard map; Herak et al., 2011). Epicenters are distributed along the strike of the system of faults in the northwest–southeast direction. The strongest earthquakes listed in CEC are the ones of 17 December 1750 (I0=VII°–VIII° Medvedev–Sponheuer–Karnik [MSK]), 22 April 1776 (I0=VII° MSK), 10 August 1838 (I0=VII° MSK), 1 March 1870 (I0=VIII° MSK), and 16 September 1904 (I0=VII° MSK). The event of 1776 probably did occur, but we were able to find only rather confusing and sketchy reports about it. According to Perrey (1850; after GDF1776), that event was most strongly felt in Bakar (Buccari; we shall refer to the places by their present Croatian and Slovenian names, with reference to alternate or older name as stated in the sources, in parentheses) where the walls of the salt‐storehouse cracked in several places. It was also strongly felt in Rijeka (Fiume) with no damage and in Trieste, Italy. However, we could find no further information about this event, which is listed in CEC after Cvijanović (1981) but on 24 April 1776, 16:36 UTC (I0=VII° MSK). The important earthquake of 1870 near the village of Klana will be elaborated in a follow‐up paper, and here we focus on three events with epicenters southeast of Rijeka, close to the town of Bakar: the large event of 1750, the one of 1838, and a somewhat smaller earthquake of 1904, none of which so far has attracted any dedicated effort to collect and interpret macroseismic data.
The study of earthquakes which occurred before the instrumental era heavily depends on the availability of reliable contemporary data on earthquake effects needed to assess the macroseismic intensity. Not being a physical parameter, intensity cannot be evaluated in a completely objective way. Nevertheless, macroseismic procedures sometimes offer reliable estimation of epicenter and magnitude (e.g., the Boxer code; Gasperini et al., 1999, 2010), and/or hypocentral depth (Musson, 1996) for historical earthquakes, provided a large enough and consistent set of intensities uniformly covering the shaken area is available. The distribution of intensity points in this region is unavoidably suboptimal, because they are limited to a narrow and densely inhabited coastal region and the islands.
Intensity estimates in the CEC are given according to the MSK‐78 scale with modifications from 1981 (Medvedev et al., 1964; Medvedev, 1978; Ad hoc Panel, 1981). Consequently, intensities estimated in this study were also assessed according to the MSK scale. We did our best to identify, retrieve, and critically examine available historical documents, reports, studies, or newspapers. We also used material from the Seismological Archives of the Department of Geophysics in Zagreb, Croatia, and the Slovenian Environment Agency in Ljubljana, Slovenia. Every effort was made to obtain and study as many primary documents (eye‐witness descriptions, contemporary accounts, or official damage reports) as possible. The secondary sources were used when the primary ones were unavailable; those were considered cum grano salis by critically evaluating the reliability of their contents.
The Earthquake of 17 December 1750
The strongest event in the 1750 series of earthquakes occurred on 17 December 1750 at about 5:30 p.m. (all times in this article are local, UTC + 1 hr, unless indicated otherwise). It was preceded by a number of foreshocks, for example, on 28/29 November around midnight (about 11 p.m. and 12 p.m.; Gratianus, 1755). According to Gratianus (1755) (whose account of this series was also extensively cited and translated from Latin to Italian, German, and Croatian by Tomsich, 1886, Müllner, 1895, and Kišpatić, 1891, respectively), in the next three years about 3000 earthquakes and brontides were felt or heard in that area. The last widely felt event occurred on 24 April 1753 (Gratianus, 1755).
We were able to consult as many as five primary sources of information regarding this event: Acta Buccarana (part relating to 1670–1776), a letter by Father (Padre) Mainardi (1751?), accounts of Gratianus (1755), the letters of military officers reproduced by von Radics (1903), and the parts of the Jesuit diary as reported in IMD1895. The secondary sources include mostly documents and papers written by compilers of earthquake‐related data (most notably Kišpatić, 1891) or of the history of those parts of Croatia (e.g., Tomsich, 1886; Laszowski, 1923; Marochino, 1978; Moravček, 2012), sometimes citing their sources of information and sometimes not. Below we summarize information from the most important sources used.
Acta Buccarana is the archive collection of official documents related to the town of Bakar (Buccari) and its surroundings, dating from 1670 to 1776. Today it is kept in the Croatian State Archives in Zagreb. The documents are an invaluable source of information about the history of these parts of Croatia, which were until 1671 the estates of the noble family of Zrin with the headquarters in Bakar. The documents, as well as the very turbulent history, are described in detail by Zmajić (1960) who wrote:
… in 1671 those estates became the property of the … Hungarian Court Chamber in Bratislava which, in 1692, sold them to the Austrian Chamber of Interior in Graz. In 1749, those estates became the property of the Bank Ministerial Chamber in Vienna and, in 1754, of the Court Council of Commerce in Vienna, which put them under the administration of the main Commerce Administration in Trieste. The coastal estates remained under its management till 1776 when the queen Maria Theresa abolished the Commercial Intendance and annexed the estates partly to the kingdom of Croatia and partly to the Croatian part of the so called Vojna Krajina, the military part of Croatia along the Turkish‐Croatian border.….
This protracted turmoil in the course of a century makes the documents rather heterogeneous in the way they were written and organized, as well as in the language used which include (handwritten) Latin, German (in Gothic and Latin script, often mixed together), Italian and, very rarely, Croatian.
Acta Buccarana was reorganized in the late 1950s (see Zmajić, 1960, for details) and are now stored in 68 boxes. The documents are in excellent condition. Although this collection was known to exist, it has, to the best of our knowledge, only been cursorily considered as the source of earthquake‐related data by R. Lopašić (see the note of personal communication in Kišpatić, 1891, p. 126) and by Laszowski (1923). Some authors (e.g., Marochino, 1978) explicitly stated that they knew of its existence and contents but had not used it. We combed through the collection, extracting the documents that were related to the damage caused to buildings and churches in Bakar and in several neighboring settlements. Here is the summary of information found:
In Bakar (Bukkari, Buccari), the cost of repair was asked from the authorities for the castle (the citadel—costs of 6490 lira), timber, salt and vegetable warehouses, the vicarage, the lighthouse, and an inn. The churches of Our Lady of the Port and St. Margaret’s were also heavily damaged. For all these buildings, there are rather detailed specifications of the quantities of various materials needed (tiles, timber, nails, iron, lime, buckets…), the manpower (masters and manual labor), and the associated costs. In Bakarac (Buccariza), repair was needed for the salt and timber magazines, the quarters of the maritime guard, and for the city gates. In Kraljevica (Porto Re), the castle and the waterfront needed a general reparation, as did the barracks and the flour magazine. In Rijeka (Fiume) the hospital, timbre office, the lazaretto, and salt warehouse are listed. The castle in Trsat (Tersat) was heavily damaged, judging from the amount of money requested (two documents asking for a total of 16,400 lira). The citadel and the warehouse in Novi Vinodolski (Novi) needed mending, as well as the warehouse in Selce (Selza) and the watermills by the sea in the bay of Žrnovnica (Sernovniza) near Novi. In Crikvenica (Zerqueniza), damage to a warehouse and to the castellan’s quarters is recorded, whereas in the bay of Sršćica (Serschiza) on Krk island a repair of the small fortress was requested. The requests for covering the cost of repairs continued for at least 13 years. In the later requests, a mention of the earthquake is usually missing. This can be due to the fact that the cause of damage was common knowledge by then, but one cannot exclude the real possibility that at least some of those requests had to do with regular maintenance of public buildings and were not related to the earthquake. It is unfortunate that earthquake damage or repair of private buildings is not mentioned in Acta Buccarana (those were considered private problems!).
We browsed through the documents pertaining to the years 1751–1763 (signature HR‐HDA‐21, series A, boxes 8, 10, 11, and 17 in the Croatian State Archive in Zagreb). We cannot claim that the search of Acta Buccarana was exhaustive and that we did not overlook some important information, because the volume of documents is quite large. It would probably be worthwhile for us to spend more time and resources, but additional engagement was outside the scope of this study and of the funding for the project.
The letter of Mainardi (1751?) was reproduced by S. Gigante in 1912, from a copy of the contemporary letter. In the letter, Father Mainardi gives a very detailed and vivid account of the effects he witnessed himself in Rijeka (Fiume). In the beginning of the letter, he describes the two foreshocks of 28 November. The second foreshock (at midnight) was very strong (“…Da una scossa sì furibonda e lunga tutti credevano che fosse l’ultima per Fiume…”), and made everyone flee and search rescue, even by boarding the boats at the sea. Feeling the boats jolting, they came back ashore. The damage to the city was not obvious from a quick glance at the buildings, but closer inspection revealed that many houses suffered some damage (“…Il danno d’oggi sofferto dalla città non fu notabile, benchè non vi fosse casa che nelle sue aperture non mostrasse segni sensibili del suo infortunio…”). On 17 December, 13 earthquakes were felt from 4 a.m. to the time of the mainshock at about 5:30 p.m. The mainshock occurred just before 6 p.m. “con furia indicibile.” The main corridor in Collegio di Fiume where Mainardi lived was impassable from rubble, the vaults and ceilings were opened up enough to insert one’s hand, the furnaces in all rooms were overturned, the walls were slanted, and all the gutters were broken down and fell to the street below. Mainardi observed that walls of many houses in the city collapsed, the stones in other walls became loose, and the tiles, parts of walls, and windows were falling down like rain (in spite of that, no one was hurt). During the earthquake, his church sustained heavy damage too (one could see the sky through cracks in the ceiling), as did the Church of the Three Kings. He also lists many specific buildings that will need rebuilding from scratch. Father Mainardi described the damage to the hill of Trsat (east of Rijeka), along with effects he experienced during the shaking (loud explosion comparable to fire of artillery, followed by shaking so severe that his lamp jumped from the middle of the table and fell down to the floor). Some perennial wells dried up, and many new ones appeared that were so rich in water that they could power a mill. The water from these new springs was turbid and colored, smelling of sulphur. Father Mainardi estimated damage to private property in the amount 147,000 fiorini (forints). To get an idea of the value of this sum of money, we note that the yearly labor cost for the construction of a 48‐m‐long pier with a small fortress and breakwater in the harbor granted by the empress Maria Theresa in 1753 amounted to 8000 fiorini (Depoli, 1921).
In the book De usu Mercurii tam externe quam interne usurpati… (Gratianus, 1755), written in Latin, the physician Gratianus from Rijeka gives a very thorough description of the complete earthquake series which started with foreshocks on 28 November that made many people flee the city and take shelter in wooden huts and tents erected at the seashore. The mainshock of 17 December lasted as long as it took to slowly say the Salutatio Angelica (Ave Maria), which is ∼15 s. Although some people were hurt, no one died; many houses sustained cracks, although none collapsed totally. After the earthquake, provisional accommodations were built outside the city. Gratianus and his family returned to their city quarters only two years after the earthquake, in November 1752, and occasionally they went back to the hut after a strong aftershock—notably after the one of 24 April 1753 at 3 p.m., which he describes as a little weaker (paulo minor) than the mainshock.
The report Geschichtliche Erinnerungen an das grosse Erdbeben in Fiume im Jahre 1750 includes several letters from military officers stationed in Rijeka in 1750 published by the historian von Radics (1903). They contain accounts of the immediate effects that the earthquake had on the buildings and people in Rijeka. The military commander of Rijeka, Chief Lieutenant von Thun, is the only one who mentions casualties (“…verschiedene Personen erschlagen…”) in his letter of 18 December. He also reported that many houses fell down or sustained cracks, and the guard houses were destroyed. The castle, the lower gate, and the bastion of St. Hieronymus were terribly cracked, and the tower in the castle fell. We also learn from these letters that two large military tents were taken to Kraljevica (Porto Re), that all clergy and nuns have left their monasteries and have gone to the suburbs, and that the municipality of Klana (Clana) asked the court for help in repairing their church, which was ruined by the earthquake.
Secondary sources indicate that parish church of St. Andrew the Apostle in Bakar fully collapsed after the earthquakes in the last days of November, and it remained a ruin for a century thereafter (Laszowski, 1923, after Acta Buccarana; Marochino, 1978). However, the belltower which had been erected in 1710 survived the earthquake. Marochino also wrote that 17 buildings in Bakar completely collapsed (including the citadel and the churches of Our Lady and St. Margaret’s) and all others were damaged (cracks and/or partial collapse). Marochino provided no reference for these data, but the primary sources do not contradict these observations. The data from the census of 1775 (Celić, 2012) show that Bakar had 1432 inhabitants and 323 buildings. As these numbers were probably similar at the time of the earthquake, a conservative estimate is that more than 5% of the buildings (mostly vulnerability class A on the MSK scale) totally collapsed, which is in agreement with our assignment of VIII° MSK in Bakar.
The City Tower in Rijeka collapsed and was restored three years later (Fig. 2). The earthquake, as is often the case, was a watershed event in the development of the city of Rijeka, because the empress Maria Theresa granted a large sum of money for restorations and approved a new general plan for urban development of the new parts of the city (civitas nova) under the condition that the Old City within the walls remain unaltered (Nepokoj, 2010).
The parish church in Grobnik was heavily damaged, as were the nearby chapels of St. Bartholomeus in Cernik and St. Marina in Majur, which were subsequently closed (Bogović, 2005). The chapel of The Holy Cross in Kačani totally collapsed (Bogović, 2005). In Belgrad, the Church of Our Lady of the Snow suffered heavy damage (Moravček, 2012). Žic et al. (2015) wrote that the earthquake caused a landslide on the right (west) bank of the Rječina river near the village of Grohovo.
The earthquakes were felt as far as Ljubljana, Zagreb, Pordenone, and Venice (∼85, 125, 165, and 175 km away from Bakar, respectively). An unnamed feuilleton author described in LWB1881 how the 1750–1751 earthquakes were felt in Ljubljana (Laibach). Based on contemporary reports from the Chamber of Commerce and the Diaries of the Jesuit College (see e.g., Latin transcript from the Diaries in IMD1895), the inhabitants strongly felt not only the mainshock but also a foreshock and a few aftershocks. Krčelić (1767?) mentioned that the earthquakes of 1750 and 1751 were felt in Zagreb and that the mainshock cracked the plaster on the ceiling of the Bishop’s residence. The Italian macroseismic database (Locati et al., 2016; see Data and Resources) lists the intensity in Pordenone as IV°–V°, probably after Baratta (1901) or the catalog by Tommasi (1888). The datum for Venice comes from WID1751, which indicates light shaking was felt there at the time of the mainshock.
The series of earthquakes that began in December 1750 consisted of several strong events. Because the mainshock came after the buildings were probably weakened by the foreshocks, and strong aftershocks could have added to the damage, it is difficult or even impossible to distinguish the effects of individual earthquakes in most localities. In particular, this pertains to evidence from the Acta Buccarana, from which we often learn something about the damage only after all of the earthquakes. Therefore, in Figure 3, we present the distribution of the 21 cumulative intensity points for this earthquake series. Ⓔ Table S1 (available in the electronic supplement to this article) presents the coordinates of places and sources of information used to assess the intensity at each locality.
As the intensities in Figure 3 are cumulative, they cannot be used to estimate macroseismic epicenters of the foreshocks, mainshock, or aftershocks. To include such events into an earthquake catalog, a representative source location must be assigned and used for the whole earthquake series. When only few intensity points are available, representative location of the source is often assumed to coincide with the locality of the highest observed intensity. In cases when at least the meisoseismal area is reasonably constrained (as is the case here), a single representative location for the whole earthquake sequence may be assessed as the barycenter of the set of observed intensities above some intensity level. This can be done by theoretical modeling of the observed cumulative macroseismic field (e.g., using algorithms like Boxer by Gasperini et al., 1999, 2010). Application of the Boxer code to the set of intensities presented in Figure 3 yielded the barycenter of cumulative intensities in the immediate vicinity of Bakar, which is close to the epicenter, as given in CEC (45.3° N, 14.5° E, after Shebalin et al., 1974). However, the intensity of IX° MSK from CEC seems to be overestimated; such an intensity would require that many buildings (20%–50%) of type A (which was prevalent there in the mid eighteenth century) suffer total collapse (damage of grade 5), which has not been reported anywhere. We therefore propose revising the maximum cumulative intensity to Imax=VIII° MSK. Table 1 presents the basic parameters for confidently identified events of this earthquake sequence.
The Earthquakes of August 1838
This earthquake series apparently started with a foreshock (hereafter, event E1, and similarly denoted for the subsequent events) on 9 August 1838, 5:46 p.m. (ADL1838), followed by a few strong events during the following night, the strongest of which (E2) was the one that occurred at 2:30 a.m. (e.g., HSOL1838) or 2:45 a.m. (e.g., ADL1838). On 10 August at about 8:30 p.m., the mainshock (E3) occurred, followed by a very strong aftershock about 10 p.m. (E4). The sources of information for this earthquake series include data collected by Kišpatić (1891), an account of Cubich (1874) about effects on the Krk island, and a number of contemporary newspaper reports.
The first foreshock (E1) is described in Rijeka as a light tremor in most of the newspapers (e.g., WIZ1838) or as strong shaking lasting for 2 s by, for example, ADL1838, and on Krk island as a forceful shake by Cubich (1874).
The collected reports agree that the foreshock in the early morning of 10 August (E2) was strongly felt in Rijeka awakening everyone (e.g., ADL1838, HSOL1838, KLZ1838, and WIZ1838). At Krk, Cubich (1874) describes only three shocks (similar to the one of the previous day) at the commencement of dawn, and which caused uneasiness.
For the two events of the evening of 10 August (E3 and E4), Kišpatić (1891) published a vivid account of a first‐hand witness, the vicar, I. Randić from Kraljevica, who reported on the effects in Rijeka where he had arrived by boat just before the first earthquake (E3). He reported that a loud subterranean rumble preceded the earthquake, a small boat was tossed upward, a large ship bounced and swung at the sea, the people hardly kept their balance, and everybody was scared and fled, crying and trying to hide. Reports from the shaken city tell us that the strong walls of the cathedral cracked, that house bells rang, and that boats were banging against each other in the port. A vault of the Trsat castle was cracked. The report in APZ1838 is the only one to mention that several houses in Rijeka, the castle, and the officers’ building in the artillery garrison were damaged and that the powder magazine (Pulverthurm) at St. Luke’s (in Zagrad, a part of Rijeka) was so severely damaged that it subsequently had to be supported by several buttresses. This damage was not specifically attributed to any of the most strongly felt events. The earthquake was strong enough that the governor, P. Kiss, hurriedly left his country villa and came to Rijeka to remain with the people until dawn (GDZ1838, WIZ1838).
The newspapers suggest that the earthquake was even stronger in Bakar and Kraljevica than in Rijeka, where it toppled several chimneys (e.g., ADL1838, the same text in INN1838 and PRZ1838). After ALL1838A the church tower in Bakar collapsed, damaging many houses (see also e.g., Perrey, 1850, or JDPL1838). No further information is known about this collapse, and it is mentioned neither in the overview of the history of Bakar by Marochino (1978) nor by Mažić (1896). We have therefore not taken this into account when assigning the intensity to Bakar for this earthquake.
Cubich (1874) gave a description of the effects of this earthquake on the Krk island (he refers only to the first of the two evening events [E3], which occurred at dusk) Many buildings in Omišalj (Castelmuschio) suffered damage, including the church and bell tower. In Vrbnik (Verbenico) cracks appeared on the walls of houses from top to bottom. The cathedral in Krk (Vegllia) cracked, where some walls were destroyed or came apart. In Baška (Besca) and in Punat (Ponte) the earthquake had passed almost unnoticed. Randić (in Kišpatić, 1891) reported that the already abandoned and ruined part of the Bishop’s palace in Kostrena collapsed. For the second earthquake (E4) of the evening of 10 August, we learn only that Randić (in Kišpatić, 1891) describes it as even stronger than the previous one (E3).
The earthquakes were felt in Trieste (e.g., ASZ1838). The aftershocks seem to have been rather weak, and they did not last very long (the last one reported in Kišpatić, 1891, occurred on 15 August, and in ALL1838B one reads that weak tremors were felt until the end of August). According to Pasarić et al. (2012) (who interpreted the account of Randić), an earthquake (apparently E3) might have caused a tsunami of intensity 2 and 3 on the Sieberg‐Ambraseys and Papadopoulos‐Imamura scales, respectively.
The bulk of the information from the reports presented above suggest that the mainshock was the earthquake of 10 August 19:30 UTC (E3) (as is presented in all catalogs that mention this earthquake series (e.g., Shebalin et al., 1974; Cvijanović, 1981; Shebalin and Leydecker, 1998; see Fig. 4). However, one must notice that some newspaper reporters (e.g., HSOL1838, KLZ1838) seem to assign certain effects (for instance, the damage to the cathedral and the Trsat castle, chimneys toppling in Bakar and Kraljevica) to the morning foreshock (E2), whereas others ascribe them to the first evening earthquake (E3), which everyone agrees was considerably stronger than E2. We believe that those newspaper articles probably present a distorted timeline of the earthquake effects, which was considered not important for the general readership.
The available evidence was not detailed enough to distinguish the effects of the events E2, E3, and particularly E4. Figure 4 therefore presents a cumulative intensity map for the whole earthquake series (see also Ⓔ Table S2), but we believe that this map mostly reflects the effects of the mainshock E3.
Event E3 is listed in three catalogs at different locations and with different estimates of epicentral intensity (ranging from VI°–VII° to VII°–VIII° MSK, see the legend in Fig. 4). Our location (using Boxer code) puts the epicenter on the Krk island, close to Omišalj, with an epicentral intensity of VII° MSK (Table 2).
The Earthquake of 16 September 1904
A moderate earthquake occurred on 16 September 1904 in the vicinity of Bakar. According to detailed witness reports collected by Kišpatić (1905), who was in charge of collecting earthquake‐related data in Croatia for the Earthquake Commission of the Yugoslav Academy of Science and Arts (with the help of A. Mohorovičić), this earthquake caused the most damage in Krasica (the trees swayed, stone walls crushed in vineyards, plaster fell, ceilings cracked, a chimney toppled, everyone felt it, those sleeping woke up, and various objects fell from shelves). In Bakar stone walls crumbled in vineyards, and in Bakarac walls cracked in some houses (accompanied by a tremendous subterranean rumble but with no major damage). In Praputnjak (Praputnik), the whole church quivered and swayed for 4 s, glasses in windows rattled, and people ran out of houses in fear, whereas in Rijeka heavy furniture moved in houses. This event was also felt in Slovenia with intensity of II°–IV° MSK, according to descriptions of effects collected by the Austrian Central Institute for Meteorology and Geodynamics in Vienna (Faidiga, 1906; Hoernes, 1906; Seidl, 1906). Those data were also interpreted by Oddone (1907), who assigned intensities according to the Cancani (1904) scale for 18 localities. The earthquake was instrumentally recorded in Rijeka at 05:37:25 (Réthly, 1906; the seismograms are lost), Ljubljana, Rocca di Papa, Padua, Florence, and Pula (Belar, 1904). Short reports about the earthquake are found in NWT1904A, GRV1904, and GRT1904, but with no details. According to Kišpatić, aftershocks were felt until 29 September, mostly in the epicentral area (Krasica, Praputnjak, Draga, and Bakar).
Figure 5 presents the intensity map for the event of 16 September 1904 (see also Ⓔ Table S3). The macroseismic epicenter computed by the Boxer algorithm is close to Bakar and coincides with the one estimated for the 1750 sequence. The maximal intensity of Imax=VI°–VII° MSK, as well as the estimated epicentral intensity of VI° MSK, are lower than the value given in the catalogs (I0=VII° MSK). Table 3 presents parameters for the mainshock and four identified aftershocks.
Analyses of available historical sources provided evidence to confidently assess macroseismic intensity at a relatively large number of localities for all studied earthquakes. This permitted us to compile, for the first time, macroseismic maps for three important earthquakes that occurred in the vicinity of the Croatian city of Rijeka in the northwest part of the External Dinarides: 17 December 1750, Imax=VIII° MSK; 10 August 1838, Imax=VII° MSK; and 16 September 1904, Imax=VI°–VII° MSK. The data reasonably constrain the epicentral locations and intensities of these events. We have also identified a number of strong foreshocks and aftershocks which are currently missing in all catalogs. Together with the Klana earthquake of 1870, these are the largest known events to have occurred in the vicinity of Rijeka (the third largest Croatian city, and the largest port), playing an important role in constraining the estimates of seismic hazard there. This role is even more important because confident identification of active faults and their properties as input to hazard assessment in the studied area is currently not possible, due to complex and still unresolved tectonic relations.
DATA AND RESOURCES
Some data were taken from the Croatian earthquake catalog (CEC) and the Croatian macroseismic database, which are the property of the Department of Geophysics, Faculty of Science, University of Zagreb. The Italian macroseismic database described by Locati et al. (2016) was queried at http://emidius.mi.ingv.it/CPTI15-DBMI15/query_eq/ (last accessed January 2017). The European Database of Seismogenic Faults (EDSF) compiled in the framework of the Project Seismic Harmonization in Europe (SHARE; Basili et al., 2013) was consulted online at http://diss.rm.ingv.it/share-edsf/ (last accessed November 2016).
This study has been fully supported by the Croatian Science Foundation, Grant HRZZ IP‐2014‐09‐9666. We gratefully acknowledge the remarks of John Ebel that improved the quality of the article. We also thank Iva Vrkić who collected much of the archive and library material, and helped with translation from Italian and Latin. Mira Pasarić kindly shared collected copies of the newspapers with data for the earthquake of 1838.