Kepler

Data sample times slowly increase, e.g. second by second, over time, instead of staying on expected markers. This makes for irregular, messy datasets. I would guess the problem lies with SPAN.

A dataset available here shows examples of this:
http://dev2.nceas.ucsb.edu/knb/metacat?action=read&qformat=default&sessionid=&docid=doc.1345243176506661893.1&displaymodule=entity&entitytype=dataTable&entityindex=1

Some example problematic changes found within:

timestamps data
2012-08-16 00:10:09 13.302116394
2012-08-16 00:10:39 13.302242279
[SNIP]
2012-08-16 02:46:09 13.295635223
2012-08-16 02:46:39 13.295513153
2012-08-16 02:47:10 13.295863152 # <== slip
2012-08-16 02:47:40 13.296142578
[SNIP]
2012-08-16 08:09:10 13.290791512
2012-08-16 08:09:40 13.290988922
2012-08-16 08:10:11 13.290698051 # <== slip
2012-08-16 08:10:41 13.290913582
...