Assessing the coverage of various collections?

Question

This is perhaps a little more of a meta-question, but it has come to mind repeatedly over the past few weeks. I'll start with a specific example.

In this question, I bring up the case of my ancestor, John White. He is claimed to be (and claims to be) from Pomfret, Connecticut. Pomfret is one of the town which is represented in the Barbour Collection, which purports to document all births, marriages, and deaths in certain cities in CT before 1850.

Here are the relevant facts:

• John's birthdate is given as October 20, 1758 in Pomfret, Connecticut on his headstone.

• There is no John White with that birthdate in the Barbour Collection.

• There is a John White in the collection with a birthdate of October 21, 1762. This is the only John White born in Pomfret in the collection and there is only one other John White born within three years of the birthdate on my ancestor's headstone.

From these facts it is clear that either my ancestor was missed in the collection or that the collection is mistaken about his birthdate or that the birthdate given on his headstone is wrong. I know that the way to resolve this is to do a little more research and look for supporting evidence one way or the other. This question is not about how to resolve this particular issue.

Instead, I have a slightly more general question. If we are given a collection which purports to be "complete" in the sense that it claims to list all of the events of a given type in a given locality over a given timeframe, how do we go about assessing the accuracy of that statement? Going back to John White, is it more likely that he was completely missed, that the collection was wrong, or that his headstone is wrong? With what frequency are there mistakes in the collection? With what frequency are there misses in the collection?

An additional bit of info on this particular case. John White was a Revolutionary veteran and is in the DAR database with the 1758 birthdate. I don't know how strictly accurate the DAR database is, but I believe that they have genealogists that determine the accuracy of lineages presented in applications, and it wouldn't surprise me if their info on actual participants was held to a higher level of accuracy than some of the other info.

Answer 1

The Barbour collection is a good example of why researchers need to be aware of exactly what and how they are searching when they investigate this type of compiled collection.

One approach is to start with a citation of an item in an online collection and track the search result back to the original source records, as described by Elizabeth Shown Mills in her QuickLesson 12: Chasing an Online Record into Its Rabbit Hole.

But since the point of the question here is that an expected record was not found, it might be more appropriate to approach the problem from the opposite direction, and ask How can I determine what records are available in a particular locale? My checklist looks like this:

1. Learn what records might have been created in a particular time and place.
2. Research which of those records might still exist.
3. Research what repositories might hold those records.
4. Research which online repositories might hold those records.

This basic list can be refined by adding steps such as Research why the records were created (usually in response to some act or law requiring the records to be collected), Reseach how the records were collected (understanding the mechanics of the process can reveal how the recording and collecting process might have gone awry) and Research how the compilation was made.

The FamilySearch Wiki's article Connecticut Vital Records gives an overview of what records were created. Statewide recording began in 1897; General Compliance did not come about until 1915. For earlier records, one has to go to the town level, where the earliest records are from 1630's (or from the date of the town's creation, if the town is newer). The main article from FamilySearch Wiki article says about the Barbour collection:

Most towns are included. The index is incomplete, however, and is known to contain errors.

The articles on the Barbour Collection at the FamilySearch Wiki and the Connecticut State Library explain how the collection came into being. The first thing to consider is that for any project of this kind, the compiler may have the goal of assembling a complete collection, but the coverage depends on the compiler's finding skill. Another issue is the form that the collection takes, which may affect how a user can get at the records. The Connecticut State Library says the transcripts were

typed onto printed forms. These form sheets were then cut, producing 12 small slips from each sheet. The slips for most towns were then alphabetized and the information was typed a second time on large sheets of rag paper, which were bound into a separate volume for each town. The slips for all towns were then interfiled, forming a statewide alphabetized slip index/abstract of most surviving town vital records to ca. 1850. Thus, there are two parts of the Barbour Collection: the slip index, and bound volumes for individual towns.

(emphasis mine)

So we have the original town books (which may not be complete) that were transcribed (by hand) and then typed for the slip files -- and then to make the printed books, typed again. Whenever you have transcriptions or copying, errors can creep in, even with the most diligent transcribers and indexers.

Note too the section on the Connecticut State Library's page that discusses the subtle differences between the slip files and the bound volumes.

The main article on the Family Search Wiki says

On 98 microfilms FHL films starting with film 002887 and the births are indexed in the IGI.

The FamilySearch catalog record Barbour collection : Connecticut vital records prior to 1850 says that the microfilms are of "original records at the State Library Hartford, Connecticut" which would seem to be the slip file. The bound volumes are also available at the Family History Library and their catalog entry is here.

For any online vendor's collection of these materials, the question must be, is their publication or index of the Barbour collection complete (i.e. do they have a subset of Barbour's subset). On the NEHGS site under search tips it says:

This database currently contains records for 124 towns. There are a total of 136 towns covered by the Barbour Collection. The remaining towns will be added to the database in groups of four throughout the year.

Ancestry.com's About this database cites the 55 bound volumes, so any of the cautions made about the bound volumes such as the regularized spellings would also apply to their database. In addition to that, if their computerized index of any bound volume is generated by OCR (Optical Character Recognition), it is possible that the name that you seek did not get rendered properly by the OCR software. (Example: In some pages of the US City Directory collections, I have observed half pages for which NO index line exists, and half pages where names were not indexed with the proper surname due to parsing errors.)

Jane Devlin's site has BARBOUR COLLECTION for CT TOWNS: TIPS that address problems with searching the collections.

Users should also be aware that online extractions made by users from either the bound volumes or the slip files will have their own limitations. E.g. one page I found had extractions for dates later than 1800 so the search for the person listed in the question from the 1750s would not be found in such an abstract.

When searches fail, it is important to remember that absence of evidence is not evidence of absence. Searching for a location without any name can be a quick way to determine if a particular town is listed in a collection or not. Ultimately if you want to chase the record all the way back into the rabbit hole, you have to leave behind the finding aids or other compiled sources of information and go back to the original town records from which the compilation was made.

Walking through the principles in Mills' Evidence Analysis Process Map can help users gain new perspective on what they do find online. Her 'rabbit hole' lesson and Do You "Just Trust" Citations Offered by Digital Providers? (posted 25 June 2014) and her followup from 06/26/2014 . . . EE's Take on the Subject can also reveal how the search can go awry, if the user considers what glitches might happen along the path back to the original record.