Time marks are superimposed on the data during recording and are subsequently removed during the initial DR100 processing. Two records; 0171408R - C8A and 0202340P - C8A failed the automatic time mark removal process. In each case a time mark near 2.1 - 2.2 seconds has been partially removed "manually" to allow integration and plotting of the traces. Part of the time mark was intentionally left to remind the reader that these records have been processed differently; the resulting integrated traces are unreliable near 2.1 - 2.2 seconds. Some steps and spikes in other records may be due to the inability of the software to perfectly remove the time marks.

The co-located velocity and acceleration instruments provided a unique opportunity to compare the responses of both instruments at high frequencies, which are especially important in the New Brunswick data set. Spectral ratio tests from the two instruments indicate that they respond as expected up to 50 Hz but deviate above 50 Hz. Lacking detailed knowledge of instrument response, we propose, as a general rule, that the New Brunswick data are reliable up to 50 Hz. The anti-alias filter serves as a low-pass filter at 50 Hz (70 Hz for C9V) and there has been no attempt to deconvolve it during the processing.

The choice of 2 Hz as a lowest reliable frequency is a compromise based on signal and noise spectra and trial-and-error processing. Some of the low amplitude accelerograms cannot be reliably integrated since they are contaminated by noise at frequencies greater than 2 Hz. Higher-amplitude accelerograms do not suffer this problem (see, for example, 0171333T - C7A). Choosing a higher frequency corner for the high-pass filter would have discriminated against the velocity records which can be realiably integrated using 2 Hz.

In summary, we feel that the processed New Brunswick data are reliable between 2 Hz and 50 Hz, except when low-amplitude accelerograms are contaminated by noise at frequencies above 2 Hz and except for possible contamination of low-amplitude records by imperfect time-mark removal. Since the velocity transducer and FBA are each flat from 2 Hz to 50 Hz, the instrument response has not been deconvolved from the records although gain factors have been removed so that the seismograms in the appendix are in units of ground motion. Each seismogram is named for the time and instrument. For example, record 0152037J - C7A was recorded at C7A on day 015, hour 20, minute 37, between 27 and 30 seconds.