Whereas the 1811-1812 sequence occurred at a time when westward expansion had just barely begun, the population of the central and eastern United States increased dramatically over the following three decades [ref]. The intensity data bases for the other two significant New Madrid events of the 19th century-the 1843 Marked Tree, Arkansas and the 1895 Charleston, Missouri events-are therefore considerably more extensive than for the earlier events. From the considerably more ``fleshed-out'' intensity distributions (Figure 3 - 1895 intensities ; ref), Johnston (1996) estimated of 6.3 and 6.6 for these events. Given the much larger intensity data sets, these estimates are considered significantly better resolved than the magnitude estimates for the 1811-1812 mainshocks. I will therefore assume the magnitudes to be correct and use the pattern of intensities to draw general inferences about site response.
The effect of amplification along major river valleys is clearly evident in Figure 3. To quantify the magnitude of the amplifications, we use [....] to determine the equivalent circular radius of MMI 4-7 shaking for 6.3 and 6.6 events. We then consider the regions where actual shaking was stronger than predicted (Figure 3). While MMI values can be amplified by other factors, such as directivity, I conclude that the pattern of amplification evident in Figure 3 is strikingly consistent with expectations for site response. Along the Ohio River Valley and the Mississippi River, wide swaths of significant amplification (1-2 MMI units) are observed.
Substantial amplification is also suggested along the Atlantic coast, in South Carolina especially, but the data are too sparse to allow quantification. For the 1811-1812 data especially, significant site response is suggested in towns like Charleston, SC, but there are no original sources that explicitly comment on the effect.