The only thing I can think of is at the time you created the second tx the unspent output used WAS CONFIRMED. Possibly there was a short re-org in the blockchain AFTER the second tx was broadcast which made the first tx unconfirmed. Then because of high number of free tx and miners including a limited number you ended up with the first tx not confirming for a while and that led to the second tx not confirming for a while.
That sounds plausible, but how often would you expect such a reorganization to occur? I have just dug through the logs and found another example: