Sorry for using vsnet-alert, but this is an important issue (for users of VSNET data to use or not to use his observations).
I checked the intervals of his observations on July 16 and found that the great majority (more than 70%) of the intervals were 0.0007 day (I initially thought that they were spaced by 1 min, but were not; exactly 0.0007 day). The cases of interval=0 were the second most abundant. So he apparently automatically created the times of observations by adding a constant of 0.0007 day. Any user would be able to confirm these statistics.
vsnet-alert@ooruri.kusastro.kyoto-u.ac.jp