Reply to “ Comment on ‘ Revisiting the 1872 Owens Valley , California , Earthquake ’ by Susan E . Hough and Kate Hutton ” by William H . Bakun by Susan E . Hough and Kate Hutton

Bakun (2009) argues that the conclusions of Hough and Hutton (2008) are wrong because the study failed to take into account the Sierra Nevada attenuation model of Bakun (2006). In particular, Bakun (2009) argues that propagation effects can explain the relatively high intensities generated by the 1872 Owens Valley earthquake. Using an intensity attenuation model that attempts to account for attenuation through the Sierra Nevada, Bakun (2006) infers the magnitude estimate (Mw 7.4–7.5) that is currently accepted by National Earthquake Information Center (NEIC). We do acknowledge and apologize for one error in Hough and Hutton (2008): the manuscript was not, as the acknowledgments state, reviewed by Bakun prior to or after submission. The acknowledgments should have thanked Bakun for helpful discussions about the inversion approach and for providing the inversion code. We were also remiss in not thanking Nancy King, who did review the manuscript. Before considering the issues raised by Bakun (2009), one needs to first examine the intensity values on which Bakun (2006) and Hough and Hutton (2008) (hereinafter B06 and HH08, respectively) are based, as any analysis of macroseismic data will rest critically on the quantity and quality of intensity assignments. As many authors have pointed out, intensity values are not data but rather interpretations, and any number of issues can be lurking within these interpretations. If these issues are not considered carefully, results can be significantly biased (e.g., Hough et al., 2000). In their investigation, HH08 revisit all known accounts of the 1872 earthquake, including those from the extensive archival search of Toppozada et al. (1981), as well as a few additional accounts unearthed by their own archival research. Accounts were analyzed with careful consideration of historical building styles, as well as with insights gleaned from macroseismic observations of large earthquakes in recent times. No intensities were assigned to accounts that describe only effects, for example, rockfalls, which are known to be unreliable indicators of intensity (e.g., Hough and Elliott, 2004). This reinterpretation yielded 142 intensity assignments, presented in the electronic supplement to HH08. In contrast, it is not entirely clear which, or how many, intensity values were used by B06. Bakun (2009) states that B06 omits modified Mercalli intensity (MMI) assignments from Toppozada et al. (1981) that were based on unreliable intensity indicators, but this is neither mentioned nor explained by B06. B06 do explain that all intensity assignments of V and below are discarded, arguing that the observations are not complete. The notion that fewer intensity values are better than more intensity values for constraining magnitude is, we suggest, a curious one. On the other hand, presumably it goes without saying that, in any scientific analysis, good data (or good intensity values) are better than bad data. B06 defends the decision to truncate MMI values at V by pointing out the MMI V values assigned by Toppozada et al. (1981) extend to the California coastline from San Francisco to San Diego, so “the truncation of MMI V values at the California coast would bias estimates ofMi if the MMI Vassignments in California were concluded.” As shown by HH08, the intensity distributions of the 1872 and 1906 earthquakes diverge most notably at regional distances, in part because of the well-established difficulty in distinguishing high (MMI > VIII) values for historical earthquakes. A comparison of the two events therefore hinges most critically on the intensity distribution at regional distances. The (unknown) number of intensity values that are analyzed by B06 are apparently taken directly from Toppozada et al. (1981), without critical examination or any attempt at reinterpretation beyond (apparently) omitting values based solely on unreliable indicators. This alone renders any conclusions about magnitude suspect, as HH08 show that a careful consideration of original accounts results in significant revisions to the MMI assignments of Toppozada et al. (1981). The conclusion reached by Bakun (2006) that “there are almost no MMI < V assignments because most people were asleep when the 2:30 a.m. event occurred” is simply wrong. There are no MMI < V assignments in the assessment of Toppozada et al. (1981) because the interpretation of accounts in this study followed practices that are now recognized to be flawed. Again, considering original accounts carefully, one finds that any number of accounts specify whether “few,” “many,” “most,” or “all” were awakened, a key distinction between MMI II–IV for earthquakes at night. Furthermore, while Toppozada et al. (1981) assigned MMI V for locations where “many” or “most” sleepers were awakened, if one looks to intensities determined from the