The Early Modern Legacy in the Modern Era
The 1703 Genroku earthquake was more powerful and more deadly than the Ansei Edo earthquake, but its long-term impact on society was minimal. Indeed, by 1855 writers had to remind Edo's residents that the 1703 event had taken place. By contrast, the social impact of the Ansei Edo earthquake was substantial and long lasting. Ansei Edo figured prominently in Meijiera debates on architecture and in debates regarding the characteristics of Japan as a nation. Major earthquakes in 1891, 1923, and 1995 stimulated reflection on 1855 in academic and journalistic circles, and its “lessons” partially informed reactions to these later events. For example, Ōmori was in Australia in 1923 when the Great Kantō Earthquake struck. In an interview there, he dismissed reports of as many as one hundred thousand deaths in the Tokyo area. Ōmori's reasoning was that similar figures had circulated in the wake of the Ansei Edo earthquake, despite an actual death toll that was closer to seven thousand.
The Ansei Edo earthquake itself was part of a longer development of earthquake culture that had been under way since the sixteenth century, some aspects of which carried forward. To take a specific example, flashes of light were part of the description of every major earthquake since 1662, and atmospheric phenomena soon became part of discussions of earthquake precursors. The Ansei Edo earthquake reinforced much of this lore and added new items such as claims about magnetic disturbances and catfish, as discussed in chapter 2. Japanese seismologists in the modern era tended to take seriously much of this accumulated earthquake lore and to devote resources to conducting research informed by it. In the paragraphs that follow, I discuss aspects of this earthquake culture, looking backward and forward from the standpoint of 1855.
An article appeared in the Yomiuri shinbun in 1915 based on a recent presentation by Imamura. It begins with the popular notion that earthquakes occur in sixty-year cycles, based on the old zodiac cycle. As reported in the article, Imamura confirms that there is some basis to this idea, but he does so by means of a statistical analysis of past earthquakes. Moreover, volcanic eruptions also tend also to follow a sixty-year cycle. Reminiscent of the idea of trapped yang energy beneath the earth, the article explains that major volcanic activity keeps earthquakes mild and vice versa. Moreover, the relative prominence of each type of geological activity alternates, thus producing something more like a 120-year cycle. Because the last major volcanic activity was in the Tenmei era (1781–1789), any earthquakes occurring around 1915 should be mild. According to this article, science has apparently verified the basic truth of lore originally based on cylindrical cycles. From the standpoint of contemporary science, the relationship between volcanic activity and earthquakes, if any, is unclear. The tendency to assume statistical regularity for seismic events continues to this day in what are referred to as “characteristic earthquakes.”
A particularly significant figure with respect to the modern influence of Tokugawa-period knowledge and lore is Musha Kinkichi (1891–1962). Musha was a seismologist who invested considerable time and energy into compiling historical documents connected with earthquakes, following up on similar efforts by Sekiya Seikei and Ōmori. Musha was a representative of the “Ōmori seismology” that lost favor after 1923. Until the end of his life, Musha argued that researchers should take seriously and further investigate certain phenomena well attested in early modern documents. He was also convinced that with more effort, earthquakes could be both predicted and prevented. For example, he thought it difficult but possible to figure out a way to release energy (enerugii) that had built up under the earth's crust and posed an earthquake danger.
In chapter 2 we saw that early modern earthquake signs included unseasonably hot weather; thunder; well water increasing, decreasing, or becoming muddy; stars appearing closer, brighter, or reddish in color; flashes of light; and, after 1855, strange behavior of catfish. Musha was not persuaded that all of these phenomena were worthy of attention. He seemed convinced, however, that when there was overwhelming documentary record of a phenomenon, to ignore it was tantamount to accusing our ancestors of making up lies. Today psychologists are well aware of the phenomenon whereby people think they see what they expect to see. As Thomas Gilovich explains, “When examining evidence relevant to a given belief, people are inclined to see what they expect to see, and conclude what they expect to conclude. Information that is consistent with our pre-existing beliefs is often accepted at face value, whereas evidence that contradicts them is critically scrutinized and discounted.” Closely related to this tendency for people “to see in a body of evidence what they expect to see” is what psychologist Kikuchi Satoru has called “mistaken correlation” (sakugo sōkan). For some people with preconceived notions about the nature of certain events, the occurrence of that event suddenly makes ordinary, mundane phenomena seem significant in hindsight. In extreme cases, people can manufacture earthquake “precursors” in the wake of happenings vaguely remembered. In his writings, Musha seems entirely unaware of the possibility that people expected to see certain things in earthquakes and therefore thought that indeed they had seen these things. To take one example that remains influential to this day, Musha was convinced that fish behavior could help predict earthquakes. Here I examine this matter in some detail as a case study of the process whereby early modern lore has influenced modern and contemporary scientific research and popular thought about earthquakes.
The basic idea was that certain fish might be unusually sensitive to some change that occurs just before an earthquake. The most common line of reasoning assumes that there must be some kind of electromagnetic change just prior to an earthquake. This point is unproven but commonly supposed, often with the alleged magnetic disturbances preceding Ansei Edo cited as evidence. Another possibility was that catfish are sensitive to vibrations from seismic waves too faint to be measured by instruments, which might precede the main shock of an earthquake. Research conducted during the 1920s and 1930s and between 1977 and 1993 has suggested that indeed, catfish are especially sensitive to electrical currents and vibrations. An April 1, 1932, Yomiuri shinbun article announced with fanfare that Tōhoku University professor Hatai Shinkishi had demonstrated that when catfish swim in a certain way, an earthquake occurs within twelve hours and that the fish in his lab had predicted almost a hundred earthquakes. One problem, however, is that small earthquakes occur very frequently, so it was unclear whether the catfish really were sensing earthquakes or whether their swimming and the earthquakes were coincidental. Nearly forty-five years later, a February 1977 headline in the same newspaper read, “Leave Earthquakes to Catfish” (Jishin nara namazu ni makaseru). The article explained a government-funded project to place both American and Japanese catfish in aquariums to serve as “an advance guard of earthquake prediction” constantly on duty. The basic setup was a device in the tank that would detect movement and sound an alarm. At a total cost of 120 million yen (US$1.5 million at the July 2012 exchange rate), the program continued until March 1993. Its advocates claimed a 30 percent “success” rate, though not in a strict sense of specifying in advance when, where, and how strong an impending earthquake would be. Instead, the standard was that the fish exhibited odd behavior within ten hours of an M3 or higher earthquake. So far, nobody has created a practical scheme whereby fish might usefully predict earthquakes. In summarizing the scientific data on catfish, Rikitake Tsuneji wrote circa 1995 that the idea of fish as earthquake predictors is “not absurd” (baka ni shita mono de wa naku) and should be considered in future research on earthquake prediction. At the time the catfish research was shut down in 1993, Rikitake said, “In ancient times and the present, east and west, people have reported such things as dogs barking before earthquakes. Even if only one in a hundred such reports are accurate, we should pursue them.”
Musha enthusiastically praised the work of Hatai and the ichthyologist Suehiro Yasuo of the Imperial Fisheries Experimental Station. As for Hatai's work with catfish, Musha claimed the correlation between the fish becoming agitated and the occurrence of an earthquake within fifteen hours or less (eight hours was typical in the case of the strongest earthquakes) was so strong that it could not have been attributable to chance. The most likely mechanism linking the catfish and earthquakes was a subtle change in voltage, and catfish can detect a change in electrical current as small as one microampere. Therefore, Musha concluded, the folkloric connection between earthquakes and catfish was not baseless superstition. Despite his familiarity with historical records, Musha did not mention that the specific idea that catfish could usefully predict earthquakes came from a tale of unknown provenance in 1855 and does not seem to have been prevalent earlier.
Expanding his discussion, Musha pointed out that both Sekiya and Ōmori conducted research on pheasants and earthquakes (pheasants sometimes substituted for catfish in early modern lore) but that fish have not received adequate academic attention. John Milne published an essay on earthquakes and lower animals in 1888, but it took over forty years for Imamura to write about the subject, even then barely mentioning fish. The publication of Hatai's findings, Suehiro Yasuo's research in the wake of the 1933 Shōwa Sanriku earthquake, and an article on the strange behavior of fish in the wake of the Oga earthquake of 1939 finally brought fish into the academic spotlight. As a whole, this research reveals four types of behavior that fish might exhibit prior to an earthquake: (1) strange behavior; (2) gathering in a group at the water's surface and frequently jumping; (3) gathering near rocks; and (4) certain species temporarily disappearing.
The following sample includes only a few of Musha's many examples. Just before the Nōbi earthquake, large numbers of loaches appeared in the wet fields of Gakuden Village in Aichi Prefecture. Before the Sanriku tsunami of 1896, vast numbers of eels gathered along the coast, and even in the daytime, normally secretive eels protruded from their holes. Prior to the Great Kantō Earthquake there were reports from the waters off the Izu coast that fish called shige (probably suketōdara [Alaska pollock], Theragra chalcogramma), which normally inhabit deep waters, could be found floating at the surface. Moreover, the Shōwa Sanriku earthquake occurred around 2:30 a.m., but Suehiro discovered that four and a half hours later at a beach near Odawara, fishermen caught a rare ribbonlike threadfish (Nemichthys avocetta) that normally lives only in waters three thousand meters deep. Suehiro concluded that the fish must have sensed some change in the focal area of the earthquake and fled at great speed, reaching the waters of Kanagawa Prefecture by around 7:00 that morning. Moreover, the stomach contents of sardines caught the day before the earthquake contained plankton normally found only at the bottom of the sea. Therefore, even plankton rose upward, perhaps in an attempt to get away from the earthquake. Suehiro, later known as “Dr. Fish” (O-Sakana Hakase), was cited in the 1977 Yomiuri article on catfish claiming that unusual fish behavior was linked with earthquakes.
Musha and others claim that prior to the 1896 tsunami many sardines gathered along the Sanriku coast, a phenomenon also reported prior to a tsunami in 1856. Moreover, in 1856, 1896, and 1933, fishermen along the Sanriku coast pulled in vast quantities of sardines just before the earthquakes and tsunamis struck, and in the aftermath of the tsunamis, they caught unusually large quantities of squid. Conversely, in 1896 just prior to the tsunami, fishermen could catch no cod or sharks, and in 1933, sea cucumbers became almost impossible to find just before the earthquake and tsunami. Moreover, prior to the Great Kantō Earthquake, fish stopped biting in Sagami Bay.
Speculating on the reasons for this wide range of strange fish behavior, Musha proposed that factors beyond changes in electric currents must be at play. He quoted Imamura theorizing that fish can detect precursor earthquakes too small to be detectable even by instruments. Upwellings of subterranean water or even oil might also play some role. Musha reported that seismologist Terada Torahiko (1878–1935) proposed three possibilities: (1) earthquake motion produces machinelike stimulation that fish can sense; (2) earthquake movement has an effect on organisms like plankton that serve as food for fish and thus indirectly affects fish behavior; and (3) earthquakes cause changes in water chemistry near the coast that affect the behavior of fish and plankton.
Following this discussion, Musha says of fish and earthquakes, “Without experimental proof, however, it is impossible to convince everyone. It is regrettable that as of now, we cannot go beyond the bounds of speculation.” Presumably, the correlation that Musha found in Hatai's experiments, allegedly too high to be attributable to chance, was not sufficient to convince many other scientists. In light of Musha's statement, it is worth considering a point from the psychological literature: “We humans seem to be extremely good at generating ideas, theories, and explanations that have the ring of plausibility. We may be relatively deficient, however, in evaluating and testing our ideas once they are formed. One of the biggest impediments to doing so is our failure to realize that when we do not precisely specify the kind of evidence that will count as support for our position, we can end up 'detecting' too much evidence for our preconceptions.” Suehiro's threadfish, for example, could have come from anywhere. The idea that it managed to swim all the way from off the Sanriku coast to a beach south of Tokyo in a few hours could only seem to be plausible evidence to someone already convinced that fish can anticipate earthquakes and that they behave much like humans would—that is, they try to escape the zone of shaking. Similarly, consider the logic in theorizing Hatai's catfish. How did anyone know that a change in electrical current precedes earthquakes? This point has never been established in a rigorous scientific manner. The idea fit with older lore about magnets and electricity, but if instruments of the time could have measured it, there would have been no need for the fish. Musha and others display a circular reasoning. They started with the axiom that fish can detect earthquakes. If so, there must be a reason, and we know that catfish are sensitive to electricity because they use a sonarlike system to locate prey. The speculation about electrical changes in the earth or subtle precursor seismic waves served to explain what we already know about catfish, namely their ability to sense earthquakes. This process has generated so much discussion of fish and earthquakes that its sheer quantity is sufficient for many people to think that there must be something to the idea.
Musha's discussion goes on to include changes in well water, which might also explain fish behavior, and the sensitivities of rats, cats, dogs, monkeys, horses, donkeys, various birds, snakes, and frogs to seismic activities. In the course of summarizing the vast quantity of reports regarding the behavior of fish and other animals, Musha returns to the same basic point he made in connection with other alleged precursors not taken sufficiently seriously by the scientific community: “It is hardly the case that both Japanese and foreigners consulted with each other and wrote down lies. It is hardly reasonable to regard every one of the extensive documents presented here as falsifications.” It is useful here to consider the issue of large quantities of poor-quality evidence. In the context of discussing the widespread popular belief in extrasensory perception (ESP), a belief similar in many ways to earthquake precursors, Gilovich explains, “Much of the evidence may be fraudulent or faulty, but there is so much of it. Can't we conclude that 'where there's smoke, there must be a fire?' We cannot, of course. . . . But this reasoning is seductive nonetheless. Indeed, it is not just the average person who falls prey to this fallacy. Well-trained scientists have been known to make the same argument.” The fish stories and other tales that came out of the Ansei Edo earthquake made “sense” in their original context. Amazing earthquake tales helped sell books like Ansei Record and Ansei Chronicle, originally commercial publications for mass audiences that have now acquired the status of venerable historical records. Moreover, early modern lore about weather patterns and other precursors derived from a conception of earthquakes that no scientists and very few nonscientists held by the turn of the twentieth century.
The topic of animal behavior and earthquakes seems to come up in the popular media in the wake of every major earthquake, but the experimental proof for which Musha longed has not been forthcoming. Tales of aquatic creatures and earthquakes continue to circulate in the mass media, even if most scientists have lost interest. We have seen in the introduction that the Great East Japan Earthquake of 2011 has generated a new round of aquatic tales and even prompted the City of Susaki seriously to consider relying on animal behavior as a defense against earthquakes. No doubt, this phenomenon of revisiting old lore is in part the result of frustration at the inability of seismologists to predict earthquakes, particularly in Japan, where law and government policy explicitly state that scientists must endeavor to predict earthquakes, supported by taxpayer money. Both the 1995 earthquake in Kobe and the 2011 disaster occurred without the slightest prior warning, despite what is probably the most concentrated array of earthquake detection and measuring devices in the world. Public expectations of science are often unrealistic, and just like a patient diagnosed with an incurable disease, there is an understandable tendency to seek alternatives when science cannot provide clear answers. In the case of earthquakes in Japan, the large accumulation of lore from the early modern era provides a rich storehouse of what we might call “alternative seismology.”
-  See Clancey, Earthquake Nation, 221.
-  For a list of precursors (zenchō) in one account, see “Ōjishin ōkaze kenmonki,” in NJS, vol. 5, supplement 2, part 1, 537–538.
-  Clancey, Earthquake Nation, 152–153.
-  “Tsūzoku jishin monogatari (jō),” Yomiuri shinbun, November 19, 1915, morning edition, 5.
-  According to seismologist Shimamura Hideki, contemporary science has been unable to shed any light on a possible relationship between volcanoes and earthquakes. See Nihonjin ga shiritai jishin no gimon 66: Jishin ga ōi Nihon dakarakoso chishiki no sonae mo wasurezuni! (Sofutobanku kurieitibu, 2008), 90–92.
-  Musha, Jishin namazu, 150.
-  Thomas Gilovich, How We Know What Isn’t So: The Fallibility of Human Reason in Everyday Life (New York: Free Press, 1991, 1993), 50.
-  Gilovich, How We Know What Isn’t So, 72, and Shimamura, Jishin no gimon, 102–103. For a detailed analysis of this and related psychological phenomena, in addition to Gilovich see Kikuchi Satoru, Chōetsu genshō o naze shinjiru no ka: omoikomi o umu “taiken” no ayausa (Kōdansha, 1998).
-  Regarding sensitivity to electricity, Shimamura points out that even if catfish can sense some kind of small change that might occur prior to an earthquake, “civilization” in the form of electricity use has greatly increased background electrical noise. Geophysical sensors located even several kilometers from railway lines, for example, pick up considerable electrical noise. It is doubtful that catfish would be able to detect subtle electrical fluctuations in such a “noisy” environment. Shimamura, Jishin no gimon, 105–106.
-  “Namazu ga ugoku to kanarazu jishin ga okoru,” in Yomiuri shinbun, April 1, 1932, morning edition, 7. See also excerpts of a presentation Hatai gave on his research, “Namazu no ugoki de jishin o yochi,” in Yomiuri shinbun, October 14, 1932, morning edition, 4.
-  Robert Geller (Robaato Geraa), Nihonjin wa shiranai “Jishin yochi” no shōtai (Futabasha, 2011), 133–134.
-  For full details on this research, see Rikitake Tsuneji, “Deeta ni miru namazu to jishin,” in Miyata and Takada, Namazue, 148–156. For details on research in the 1920s, see Musha, Jishin namazu, 16–22. Regarding animals other than fish, see 23–51.
-  Quoted in Geller, Jishin yochi, 134–135. Geller points out critically that if one in a hundred is a suitable yardstick, then every superstition or fantasy ever connected with earthquakes deserves public funding.
-  Musha, Jishin namazu, 19–21. More recently, Motoji and many others have made the same claim. See Earthquakes and Animals for abundant examples.
-  Musha, Jishin namazu, 23–24.
-  Ibid., 24–29, and Yasuo Suyehiro [Suehiro], “Some Observations on the Unusual Behavior of Fishes Prior to an Earthquake,” Earthquake Research Institute, Tokyo Imperial University (March 1934), 228–231. According to Suehiro, recent studies “equally prove the existence of a mysterious and instinctive reaction of the fish to earthquakes” (228). For photos of sardines and the threadfish, see 230.
-  Quoted in Geller, Jishin yochi, 133.
-  Musha, Jishin namazu, 31–35. See also Yoshimura Akira, Sanriku kaigan ōtsunami (Bungei shunjū, 2004), 16–20, 81–82.
-  Musha, Jishin namazu, 36–40.
-  Ibid., 40. It remains a common rhetorical technique among advocates of earthquake precursors and prediction to claim that they are offering possibilities, not definite proof or precise tools. Motoji, for example, says, “I am not arguing for earthquake prediction using the animal precursors . . . if, by prediction, we are meaning an exact forecast of time, epicenter, and magnitude of an earthquake. In that case no prediction is possible for any fracture phenomena.” See Earthquakes and Animals, ix.
-  Gilovich, How We Know What Isn’t So, 58.
-  Musha, Jishin namazu, 40–50.
-  Ibid., 51.
-  Gilovich, How We Know What Isn’t So, 158.
-  This topic is beyond the scope of the present study. For a thorough analysis, see Geller, Jishin yochi, esp. 76–159, and Shimamura Hideki, “Jishin yochi” wa uso darake (Kōdansha, 2008).