Error on Census?

3 posts / 0 new
Last post
tmphelps's picture
Error on Census?

It was supposed to be easy. I was looking for the Asa Phelps household in Franklin County in the 1850 US Census. (I am using FamilySearch in this example because of the “difficulty” introduced by the way Ancestry groups and displays census records.) When I did a record search using this information, I was pointed to the following image: Being a good student of EE, I immediately examined the image and put together a citation in the form of a reference note:

1860 U.S. census, Franklin County, North Carolina, population schedule, Louisburg District, p. 768 (penned), dwelling 1013, family 1013, Asa Phelps household; imaged at FamilySearch ( : accessed 7 September 2017); citing National Archives microfilm publication M432, roll 630.

Then, since I was at the bottom of the page, I checked the next page/image to see if there were more children, which there were. That’s where I found the problem (actually I admit I originally found the problem thanks to the way Ancestry groups census records.) On this page the district is listed as "Davis District." This sent me paging forward and backwards through the Franklin County census and I saw multiple instances of this. For example at the bottom of this page a new family is started that continues on the next page where the district is labeled "Cooks District."

I understand that what we see as images are taken from microfilm of a set of records that are copies made from the census taker’s original notes. EE always stresses that our citations must state what the original record says, and if we feel we need to “correct” it we do so using square brackets. In this case, I have no idea how to go about creating a “correction” or even whether one is needed.

I hope you can offer some guidance on this.


EE's picture

Tom, you've spotted a can of worms in your thoughtful scrutiny of what you're using. In fact, we could characterize your issue as one of analysis, rather than source citation. Citing this census page is simple enough. Probing deeper suggests the issue lies in the  arrangement of the data intrinsic in the record itself--and the implications it has for your conclusions about the people.

First, by expanding our examination to a broader range of pages at FamilySearch, we see that it's not just an issue of switching back and forth between the two districts you mention (Cook's and Davis's). As an overview:

  • Image 376 (p383), Louisburg Dist., 1 October, nos. 987–994
  • Image 377 (p383verso), Cook’s Dist., 2 Oct., nos. 995-100-
  • Image 378 (p384), Perry’s Mill Dist., 2 Oct., nos. 1001-1006
  • Image 379 (p384v), Louisburg Dist., 3 Oct., nos. 1007-1013
  • Image 380 (p385), Davis Dist., 3 Oct., nos. 1014-1022
  • Image 381 (p385v), Cook’s Dist., 4 Oct, nos. 1023-1029
  • Image 382 (p386), Cook’s Dist., 5 Oct., nos. 1030-1038
  • Image 383 (p386v), Cook’s Dist., 5 Oct., nos. 1039-1045
  • Image 384 (p387), Cook’s Dist., 8 Oct., nos. 1046-1051
  • Image 385 (p387v), Perry’s Mill Dist., 8 Oct., nos. 1052 ….
  • Image 386 (p388), Perry’s Mill Dist., 10 Oct. …
  • Image 387 (p388v), Perry’s Mill Dist.15 Oct. …
  • Image 388 (p389), Perry’s Mill Dist. 15 Oct.…
  • Image 389 (p389v), Cook’s Dist. … 16 Oct. ...
  • Image 390 (p390), Perry’s Mill Dist., 17 Oct. ...
  • Image 391 (p390v), Perry’s Mill Dist., 21 Oct. ... (end of district, with some interesting notes added)

Dwelling/household numbers are all consistent. The date pattern is chronological, but the census taker is switching in and out between four different districts.

As you noted, Ancestry organizes its images of this census differently from FamilySearch.

  • FS presents the images exactly as they were filmed, which follows the sequence in which the pages were numbered and bound by the Census Bureau.
  • Ancestry takes pages out of order and presents them according to the “district” cited at the top of the page. Therefore, when a household from, say, Cook’s District is begun on one page and finished on the next page on which a new district begins, Ancestry arbitrarily splits the household between those two districts.

For citation purposes, we would cite the district name that appears on the page on which the household begins. We would do this regardless of whether we were using the Ancestry arrangement or the FamilySearch arrangement.

However, the more important issue is why the census taker created that erratic cluster of pages. For this we would normally analyze several issues, starting with

  • Who was the census taker? What can we find out about him that would explain his seemingly irrational thinking?
  • Is there a significance to the dates? What was going on, on those dates? (If, for example, a cluster like this all appeared on a Sunday, we might hypothesize that the census taker gathered data at church. Or if it were Saturday, that he gathered it at a tavern or country store. In this case, using a perpetual calendar reveals that the dates cluster tightly on week days at first, then diverge into a couple of other patterns.
  • On what days did Franklin County hold its court sessions in October 1850? Could the census taker have been gathering data there—given that court days often brought rural residents into town for celebrations of sort—rather than making rounds?

Ancestry’s rearrangement of the pages into districts suggests something else. You’ll notice that for each of the districts involved in this erratic cluster, there are a number of pages, with the pages from the above cluster being significantly out of chronological and numerical order with the rest of that district's series.

All things considered, my starting hypothesis at this point would be that the cluster of pages stamped pp. 383 through 390 verso represent cases in which the census taker went back into those districts to pick up strays that he missed when he made his first pass through each district. Perhaps these individuals were not then at home. Perhaps the weather conditions made it impossible for him to get into certain neighborhoods on his first trip.

And now, for all our readers who are wondering What does this matter?, the takeaway is this:  I’d be extremely hesitant to assign “neighbors” (and potential kin) to anybody on these pages without further study.  It would be instructive, for example, to compare each page of this 1850 cluster to the 1840 and 1860 enumerations to see if individuals had the same set of neighbors.

And then, of course, if we were using the Ancestry version, there’s that problem of those pages that begin with partial entries carried over from prior pages. For example, in Ancestry’s cluster of 15 images for Perry’s Mill, who were Calvin and David who appear at the top of image 10?

The Editor

tmphelps's picture

Thank you for the excellent analysis lesson. I was totally confused when I started to look at the situation, and your recognition of the underlying problem coupled with your study and feedback of the entire census record is a lesson that I hope I will remember and learn from for future work.

And, as you pointed out, the citation for the specific family was the easy part.