Just a guess, but I'd say right off hand that it's the fractional seconds. If a VI is running very fast it's possible for it to iterate faster than once per second. The result would be several samples with the same timestamp.
Including the fractional seconds allows the software to distinguish between closely spaced runs.
Mike...
PS: the "PM" is in a wierd place, I'd report it to NI as a bug...