This excellent paper presents valuable data and interpretation of the distribution of damage in Adapazari during 1999 Izmit (Kocaeli) earthquake in Turkey. Contrary to the traditional views that damage is greater on "poor" ground, and consistent with similar observations in California, the Philippines, and Japan, the authors [1] show that the damaged buildings and areas with large soil deformations show remarkable separation. Their observations suggest that the structural failures occurred due to strong shaking at "stiffer" sites, where ground motion was amplified, while the buildings on the "soft" sites benefited from the natural base isolation effects, due to strong non-linear response.
These observations are in good agreement with the observed trends of damage levels relative to peak ground velocity and surficial geology, during the Northridge, California, earthquake of January 17, 1994. During this earthquake, ground velocity exceeded about 170 cm/s [6], and was high (40-90 cm/s) in broad areas where damage could be correlated with the nature of the surface deposits. In Fig. 1 we show the peak velocity versus the number of damaged buildings per 1000 housing units (hu) N * . The solid points indicate adequate, the large open circles marginal, and the small circles very marginal numbers of data points, for the range of V m and N * values (Holocene: Qyf -fine-grained sediment -silt and clay; Qym -mediumgrained sediment -sand; Qyc -coarse-grained sediment -gravel; Qyvc -very coarse-grained sediment -cobbles and boulders; Pleistocene: Qof -fine-grained sediment -silt and clay; Qom -medium-grained sediment -sand; Qoccoarse-grained sediment -gravel). The solid lines show the trends through the data points for Pleistocene (Qof, Qoc and Qom) and for Holocene (Qyf, Qym, Qyc and Qyvc). It is seen that, for most of the damaged buildings on Holocene, the rate of damage growth with velocity is much slower than for the buildings situated on Pleistocene [4,5]. It will be of interest to compare these trends further with association of damage data from Adapazari, and the types of surficial geology there.
The spatial distribution of damaged buildings and of pipe breaks can further be used to determine which method of