I should emphasize that the fault here lies primarily with NARA for rebinding the censuses, but both Ancestry.com and Genealogy.com have compounded the problem by indexing and grouping the pages in their disordered state.  They could fix the problem, but it would require cutting each of the images, which are two-page spreads, in two and reassembling the individual pages in their original order, then correcting the index entries and re-grouping the images.  Short of doing that, it would be preferable for them to not break the image groups down below county and to not index them below county because no data are better than bad data, and the present situation is generating a lot of errors in the work of novices.

I have worked on disentangling some counties (linked below), based on my own research needs, but anyone using any of the 1790 through 1820 censuses should be aware of the problem and be alert to detecting when a county's census has been damaged this way.  Beginning with the 1830 census, the population schedules were pre-printed, thwarting the urge of any nitwits at NARA to have them rebound

Most census pages have multiple page numbers written or stamped on them.  Some of these numbers may have been placed on the pages while they were still in their original order, while others have been placed after the re-binding.  And here is where another incredibly incompetent practice by the Archives has compounded the problem:  that of only numbering every other page, which means half the pages contain no numbers, at all.  The practice of not numbering every page is a bad one, even when the pages aren't rebound.

It's a basic principle of data storage and retrieval that every item/specimen/page/etc. should have a unique identifier.  The lack of a unique identifier for half of the pages of some censuses has not only led to them being mis-grouped and mis-indexed, it has led to some inconsistent and idiosyncratic patterns of referring to these blank pages.  Some refer to the numbered page as, for example, "87," and the next blank page as "87A"; others refer to the numbered page as "87A" and the next blank page as "87B."  And both patterns assume the next blank page on the microfilm is truly the other side of the previous page, which it isn't if the census was rebound.  The extra work it would have taken to number both sides of the pages from the outset pales in comparison to the extra work and confusion caused by numbering only one side — but obviously, idiocy reigns at NARA!

By recording all the various numbers on the pages in a list of page images, the pattern of the numbers will reveal to anyone familiar with publishing and page imposition that the pages were rebound and how they were rebound, and this is what I've done for the censuses linked below.

The importance of making these reconstructions is that by doing so you can recognize mistakes in the indexing of a householder's location and in the grouping of the digital page images.  You may not care that you have a wrong location for your householder, but if you want to construct an accurate history of your family, you should care…

1810 New York Census:  Cayuga County

1810 Kentucky Census: Nelson County

1810 Kentucky Census: Washington County

1810 Virginia Census:  Louisa County

1810 Rhode Island Census:  Washington County

1820 Kentucky Census:  Logan County (with Index and Extraction)

1820 Ohio Census:  Knox County (with Index and Extraction)