In executing 3dDeconvolve for simple regression on an event-related experiment with 2 classes of stimuli (Pic=real pictures; PPic=pseudo-pictures), I received these messages:

** WARNING: input stimulus time series file Stim_1_Pic.1D[1] is too short:
length = 1100, but should be at least 1197.
** WARNING: input stimulus time series file Stim_2_PPic.1D[1] is too short:
length = 1100, but should be at least 1197.

Huh?

I created the stim files be executing the following waver commands, using the actual stim times based on my Presentation logfiles:
waver -dt 2.0 -GAM -xyout -peak 1 -tstim 21.7342 30.7355 (...) 2122.4703 2130.4693 -numout 1100 > Stim_1_Pic.1D
waver -dt 2.0 -GAM -xyout -peak 1 -tstim 93.736 101.728 (...) 2192.474 2199.481 -numout 1100 > Stim_2_PPic.1D

I used 1100 with the -numout option since my 8 concatenated EPI's end at almost 2200 sec and my TR is 2 seconds. Is this where I went wrong? I can't understand where the 1197 figure in the warning message comes from.