We used to have this happen a long time ago, I think in New York or Orlando.
In our case, what caused it was if an article reached its Valid To date and thus expired, and we then went in to renew it, it would skip a version number for some reason.
At this time, we had to check out articles manually and renew the Valid To date in each one individually. (IT WAS THE BIGGEST PAIN.) We have since then figured out how to turn on the permissions to give me access to update the Valid To list from list form. MUCH FASTER! (We're talking 6-10 hours depending on the month compared to 30 minutes.)
Anyway ... if your article expired before you could renew it, and then you renew it to publish it, that could be one of the issues. If you have a QA environment that you can access to test, you could purposely let some articles expire and then renew them to see if you can recreate the issue.
Have you already talked to your system admin - and have you had a system upgrade recently? My other thought is that there could be some permissions with your profile/role that got messed up somewhere along the way - for you not to be able to see the previous versions.
Or, have your admin check system properties - maybe something in the naming/counting convention of your articles has been changed.
Or, maybe when an update happened, it messed up some of your historical information. I say this because we've had it happen where when certain update sets or system upgrades have happened, it changes all the "modified on" dates to the date and time of the update, and it changes the "updated by" to the system admin name. It's frustrating, but not the end of the world. But I'm wondering if something similar has happened here.
I know it's not a full-on answer. However, maybe this gives you some places to look into at least. Good luck! Please feel free to reply with questions and let us know what you find out!
