devinemke, the problem with that is that the row doesnt even exist in the table.
I know because there are 24 hours in a day and should therefore be 24 rows per day (one observation per hour).
That's one way of looking at it.
Unfortunately, that is also one way that is difficult to translate into an algorithm, since it makes use of human pattern recognition, which can be a bit of an overkill to implement (artificial neural networks, anyone?).
If the observations are guaranteed to be at: yyyy-mm-dd HH:MM
for unique values of "HH", then a solution would be to check that "HH" is in running order.
This cant be (easily) done on the database with your current database setup, so you'll have to use PHP.
To obtain the value of HH, you can either use regular expressions, or better still, take advantage of the fixed length of the date string and get the substring corresponding to HH
After that, just implement the check by means of a counter.
You'll have to be careful about when the counter starts and ends though, since that is dependent on the range specified.