Impact on onarchive after adding chunks (7.23.UC1)

Hi, We use onarchive to do parallel backups of a 100 GB database. This process usually takes 8-9 hours. Just recently I added about 40 chunks to this environment (less than 10% additional space). The first time I executed onarchive after the addition of these chunks, the archive completed in a whopping 23 hours (14 hours more than normal). Subsequent archives fell back to the normal range (7-9 hours). Now, I realize that onarchive does not distinguish between allocated pages which are in use or not in use. In fact, it is my understanding tha that a page is considered not in use if it has been allocated but has an invalid address. I also realize that pages with "very old" time stamps will be updated by onarchive with a more current time stamp. This is done to prevent this page from being skipped in the future. The question (finaly :-) ): Assuming all of the above, plus the fact that the newly added chunks did not contain data (they were simply added to the environment but data has not made it into the chunk yet), why did my time increase as significantly as it did the first time? Why did subsequent archives execute at a normal rate? BTW, even more puzzling, this type of behavior does not exist every time chunks are added. Thanks!!!