Michael,

I don't know what to tell you, based on what you laid out in this message I think you have about the best handle on what to do that you're going to get. I also read Louise's thoughtful reply and agree with all she shared. 

Our maps cataloging team has always used the "does the item contain 50% or more maps" rule of thumb, in conjunction with looking throughout the "book" to see if the maps truly are the key element or if the majority act as illustrations for the text. And then as you note, make a decision and act upon it. 

I will say I disagree with your statement that "some" older atlases were cataloged as books/monographs -- until the late 1990s they all were considered monographs and cataloged in OCLC using the books format. And as us catalogers know, we are left to clean up that legacy. It is unfortunate that for decades institutions instructed catalogers to catalog atlases (geographic of course, don't get me started on how many times our monographs folks sent over something like an atlas of amphibians or the human body for us to catalog) as books, or essentially took the easy way out. 

In my experience I believe you will find that the majority of what you have are truly atlases, and thus when you find they were described as books you'll have to create new records to treat them properly. It will be a pain but at the same time you'll be doing many other catalogers a favor. We have diligently put in the work to convert former book records into atlas records (including using the appropriate 007 field and in at least the past 8-10 years always including geographic coordinates, as well as adding a scale statement) when going through our own atlas collection but I'm sure there are many more needing attention. Good luck on this project and unfortunately I do not believe there is any "firmer" guidance that you can turn to to resolve a particular item's appearance of book vs. atlas.

Paige

---