Lawson’s (1908) Descriptions of Shaking and Damage in the 1906 Earthquake Associated with Modified Mercalli Intensities


Not felt


Felt by people at rest, but not miners in works, lamps and open doors swing, some pendulum clocks stopped


Slight shock


Felt by most people, usually for a short duration (< 20 s), direction of motion described


Light shaking, most sleepers awakened, doors and windows rattled, longer duration (>30 s) and variability of motion described, water thrown from horse-troughs, water tanks, and canals


Moderate shaking, objects shifted, milk spilled from pans, houses rocked with slight plaster cracking, some water tanks thrown down


Trees strongly shaken, grassland and fields appear to move in waves


Heavy shaking, objects moved and thrown from shelves, plaster cracked, windows broken, some chimneys and poorly braced walls damaged, bricks thrown from parapets, tall monuments shifted


Ground cracks on roads and hillsides, some chimneys damaged


Most chimneys thrown down or damaged, some masonry but no frame buildings damaged, piles of cordwood overthrown, some headstones overturned, small landslides and earth-slumps


Liquefaction and large lateral spreads, all chimneys thrown down


Well-built masonry damaged, some frame buildings shifted on their foundations, headstones and cemetery monuments overturned, extensive ground failure and settlement, foundations, water and gas pipes broken, railway tracks twisted


Men, horses, and cattle thrown off their feet, bridges wrecked, frame buildings thrown from their foundations


Masonry and frame buildings destroyed, massive landslides, pervasive ground failure, limbs broken from healthy trees


Trees topped, almost all headstones and cemetery monuments thrown down

An Explanation of the 1906 MMI Intensity Scale

Because Lawson’s (1908) reports of the 1906 earthquake effects are so extensive, it is relatively straightforward to correlate these effects and arrange them into a self-consistent intensity scale. The process of fitting this scale to Richter’s (1958) Modified Mercalli Scale is more difficult. This difficulty stems from two sources. First, the large magnitude and long duration of the 1906 earthquake shifts the occurrence of many common shaking effects. Second, the “modern” MMI scale used by Stover and Coffman (1993) and regressed against ground motion by Wald et al. (1997) has been modified from Richter’s (1958) scale at the higher intensities. We use the modern scale but compare our intensities with Toppozada and Parke’s (1982) intensities for 227 sites.

The 1906 MMI intensity scale is very close to Richter’s (1958) scale for MMI 1-5, the so-called “felt” intensities. There are two important disagree­ments, however. As noted by Toppozada and Parke (1982), stopped (pendulum) clocks and swinging lamps and doors can occur at MMI 2 in the 1906 reports, rather than at MMI 5 as specified in Richter (1958). Often, “a clock stopt” is the only shaking effect described at a distant site. Similarly, seiches can occur at MMI 4 in the 1906 reports, rather than at MMI 7, as specified in Richter (1958). “Water thrown from horse-troughs” is a characteristic description. We presume that these shifts are the result of the enhanced long-period radiation of the 1906 earthquake.

Many of the most distant felt reports are extremely brief, often simply “a slight shock was felt.” Because we could not distinguish whether these reports correspond to MMI 2 or 3, we assigned MMI 2-3 to these sites. For sites at which the shaking was more fully described, the perceived shaking duration varied from 15 s to more than a minute. Characteristically, the shorter estimates of duration came from sites where the shaking effects were weaker. We assigned MMI 3 where the estimated durations were shorter than 20 s and MMI 4 where the estimated durations were 30 s or longer.

At the turn of the century, milk was delivered before daybreak. Milkmen would pour the fresh milk into shallow pans that had been left out in kitchens and pantries. These pans of milk functioned as MMI-5 intensity indicators, exactly as characterized in Richter’s (1958) scale. Similarly, Richter (1958) lists slight cracking of plaster, windows rattling, and small objects moved as MMI-5 effects. Taken together, these effects constitute a residential benchmark for MMI 5 in the 1906 MMI scale.

Lawson (1908) contains twelve reports of trees “lasht as tho by a gale” and eight reports of fields moving “like waves of the ocean.” Two of these reports are from the same sites. Because most of these reports come from sites where the shaking was slightly stronger than MMI 5, we use them to bound the intensity as MMI ≥ 5-6. That is, if the only effect reported at a site was “trees strongly shaken,” we would assign MMI 5-6 to that site if nearby sites with similar surficial geology had equal or lower intensities. This bound is close to Richter’s (1958) interpretation of strongly swaying trees as an MMI-6 effect.

Inside frame houses, MMI 6 is marked by objects shifted or thrown from shelves. This effect repeats the chief element of the “supermarket” intensity scale proposed by Nason (1984). In frame buildings, the rocking associated with MMI 6 shaking produces obvious non-structural damage, such as cracked plaster and broken windows. In brick buildings, MMI 6 shaking can produce cracks in poorly braced walls and dislodge bricks from parapets.

Because chimney damage can be assessed rapidly from cursory inspections of small towns and neighborhoods, it provides a critical measure for estimating intensity. Stover and Coffman (1993) reshaped the modern MMI scale so that the incidence of chimney damage spans the range from MMI intensity 6 to 8. They assign MMI 6 when one or a few chimneys are damaged or thrown down, and MMI 7 when half or more of the chimneys are thrown down. In contrast, Richter (1958) characterizes these damage states as MMI 7 and 8, respectively.

Damaged and fallen chimneys are the most common shaking effect reported for MMI 6 through MMI 8 intensities. Thus the disagreement between Richter (1958) and Stover and Coffman (1993) is critical. We find that Stover and Coffman’s (1993) interpretation is more consistent with the occurrence of other shaking effects in the Lawson reports.  In particular, many sites are described in Lawson (1908) as “all chimneys damaged,” but with no other damage to wood-frame buildings. When almost all of the chimneys are reported as damaged or thrown down, we bound the intensity as MMI ≥ 7-8. Because Toppozada and Parke (1982) use Richter’s (1958) MMI scale, our intensities are usually less than theirs for MMI 6 to 8.

One of the most well documented effects in Lawson (1908) is the occurrence of ground cracks along roads and near the tops of slopes. These ground cracks are assumed to represent soil failures or incipient landslides. We use these observations to bound the intensity as MMI ≥ 6-7. Richter (1958) assigns MMI 8 to ground cracking, but the incidence of these cracks clearly increases with the duration of shaking. The only sites where ground cracks occur near MMI < 6-7 estimates are the numerous cracks mapped by E.S. Larsen in the hills west of Round Valley (MMI 6 in Poonkinney and Covelo), and the cracks at Peach Tree Springs on the west flank of Mount Diablo (MMI 5-6 in Clayton).

Similarly, we use observations of small landslides to bound the intensity as MMI ≥ 7. This bound corresponds with the median minimum intensity discerned by Keefer (1984) for “coherent slides” generated by 40 large and great earthquakes. Richter (1958) also assigns MMI 7 to small landslides. There are no reports in Lawson (1908) of small landslides associated with intensities MMI < 7, so the bound we obtain from these reports is an impor­tant one. In general, the size and frequency of landslides increases with the intensity, so large landslides are assigned MMI ≥ 8, and massive landslides are assigned MMI ≥ 9.  We make an exception for the San Pablo Slump, however, as E.S. Larsen reported that landsliding had initiated before the earthquake occurred. 

Many marshes and low-lying estuaries were strongly shaken by the 1906 earthquake. There were a number of reports of lengths of track where the railroad bed had “disappeared” into a marsh: in the most spectacular occurrence, eight miles of Southern Pacific track subsided into the Suisun Marsh. By correlating this subsidence effect with nearby intensities, we find that it is associated with MMI ≥ 7. We note that this limit under­estimates the intensity associated with the track subsidence along the eastern shore of Tomales Bay.

Extensive liquefaction and lateral spreading effects were produced in the Eel River lowlands, and along significant lengths of the Russian River and the Salinas River. We assign MMI 7-8 to areas where there were extensive lateral spreads but no other effects we could use to estimate the intensity. This choice is predicated on the well-documented MMI 8 intensities assigned to Ferndale and Salinas, two towns surrounded by extensive lateral spreads. We assign MMI 7-8 instead of MMI 8 because we assume that the lateral spreading acts to reduce the shaking intensity, as the failure on the sub-horizontal plane decouples the surface layer and damps the shaking.

At MMI ≥ 8, ground failure effects in urban environments, particularly ground settlement, become more pronounced, breaking water and gas pipes, found­ations, and even twisting railway tracks. Many of these ground failures occur in poorly compacted fills and street grades.  In mapping the apparent inten­sity in San Francisco, H.O. Wood as­signed these ground failures inten­sity B on the San Francisco scale, which corresponds approximately to Stover and Coffman’s (1993) MMI 9.  Where these ground failures do not correspond with damage to wood-frame buildings (for example, on Union Street near Steiner or on Van Ness Avenue near Vallejo), Wood’s assign­ments appear to overestimate the intensity. 

Richter (1958) uses an elaborate scheme of weak and strong masonry buildings to assign intensities to damaged and destroyed brick buildings. Other than to distinguish adobe from stone and stone from brick, it is impossible to apply such a scheme to the Lawson (1908) reports. The reports often claim poor construction as the cause of earthquake damage to masonry buildings. We take these claims into consideration for Agnews State Hospital (brick) and Stanford University (stone). Otherwise, we assume that most brick buildings behave similarly and expect poorly braced walls, parapets, and cornices to fail at MMI ≥ 7, and brick buildings to be strongly damaged or destroyed at MMI ≥ 8.  Stone buildings are assumed to be some­what weaker than brick buildings, and adobe buildings, such as the Mission at San Juan Bautista, are assumed to be weaker still.

At the turn of the century, frame buildings were rarely attached to their foundations. At MMI ≥ 8, these structures can be damaged by sliding off their foundations. As the intensity increases to MMI 8-9, reports of frame buildings “thrown from their founda­tions” become more prevalent. We interpret these reports as indicating the failure of cripple walls. We note, however, that wood-frame houses at the turn of the century appear relatively strong, as there is a clear separation between the damage states of “all chimneys down” and “houses shifted” or “thrown” from their foundations.  Because of this separation, we assign MMI 8-9 to reports of damage or shifting of more than one wood-frame house in a small or moderate-sized community.

Robert Mallet’s (1862) study of the 1857 Neapolitan earthquake had a pronounced influence on the isoseismic investigations of the 1906 earth­quake. The throw of tombstones and monuments was meticulously docu­mented in many cemeteries. While the directions of falling monuments were more random than systematic, the damage to cemeteries can be used as an important benchmark for intensity. At MMI 7, one or two headstones can be overturned; at MMI 8, many headstones and monuments are “thrown down.” We associate reports of almost all the monuments in a cemetery being damaged or thrown down, as occurred in Tomales, Sebastopol, and Petrolia, with the highest intensity, MMI = 9-10.

Because many reports in Lawson (1908) are taken from personal accounts of the earthquake, there are numerous descriptions of people, horses, and cattle “thrown down” by the ground shaking. We interpret this effect to correspond to a number of cycles of horizontal acceleration that exceed half the acceleration of grav­ity. In Wald et al.’s (1999) correla­tion of peak ground acceleration against MMI intensity, PGA = 1/2 g is correlated with MMI 8. If we assume, how­ever, that a number of cycles of horizontal acceleration exceed or approach 1/2 g, then we expect PGA ~ 2/3 g, which correlates with MMI ≥ 8-9.

There are six reports in Lawson (1908) of tree limbs snapped and dead trees topped. Five of these sites have intensities of MMI = 8-9, so we use the effect to bound the intensity as MMI ≥ 8-9. In turn, there are six reports of live trees topped: Monroe (Hales Grove), Fort Ross, Bear Valley (Point Reyes), Ben Lomond, Summit Ridge, and Skyland Ridge. With the exception of Monroe, all of these sites are close to the fault, so it is natural to associate the highest level of shaking intensity, MMI 9-10, to these topped trees. We note that the 1989 Loma Prieta earthquake topped trees on Summit Ridge as well.

We have identified four shaking effects as indicators of the highest intensity level: massive landslides, devastated cemeteries, topped trees, and pervasive damage to wood-frame structures. We have been conservative in assigning MMI = 9-10, however, and generally require two of these effects to occur at the same locality. Because cemeteries are rarely situated on steep slopes or in wooded areas, it is difficult to combine these effects, and in turn, difficult to justify MMI > 9.

For example, the town of Petrolia suffered MMI 8-9 shaking damage, where many frame buildings were damaged by sliding from their foundations. Every monument except one in the cemetery was overturned. A large landslide occurred on the other side of the Mattole River, and a massive landslide occurred at Sea Lion Gulch, nine miles south of town on the coast. While the cemetery damage, 1 km from town, can be considered together with the damage to the town, the large landslide 3 km from town is a weaker (MMI 8) effect, and the massive landslide at Sea Lion Gulch is too far away. Thus, we can identify a relatively large area that suffered MMI 9, but we are unable to confirm MMI ≥ 9 for Petrolia itself.

Because of this conservative approach, we only assign the intensity MMI = 9-10 to four sites. In the city of Santa Rosa, both the downtown area and many wood-frame buildings were destroyed and the cemetery, situated to the northeast of town, was heavily damaged. The towns of Sebastopol and Tomales both suffered extensive damage to their wood-frame buildings and their cemeteries were almost completely devastated. The area near the San Andreas fault above Wright’s Station suffered both landslides and pervasive damage to wood-frame structures.