Thank you Geng!
Yes, that seemed to solve my problem. 3dDeconvolve went on its merry way with the new full-length vector waveforms.
While the section on the waver program in the AFNI manual had no mention of -tstim or -numout, I see now how that -numout option was included in the set of potential operators in the --help output for waver.
Inasmuch as irregular stimulus timing vectors from -tstim are ideal for button presses (which subjects may or may not make), coupled with the possibility that waver with tstim may generate modeling waveforms of insufficient length, as part of the never-ending quest for AFNI perfection, it may be a good idea to either:
1) require an explicit instruction operator in the waver command specifying number of points in the time series, or
2) include a sentence or two in either the AFNI manual and/or the waver --help subsection on -tstim cautioning how waver may not output enough timeseries points using tstim depending on when the last event is, and that -numout (see below) should be used to ensure an adequate length.
Thanks again.
Jim B