Ancestry 1931 Canadian Census Name Indexed

I woke up on Friday, 9 June to find the name-indexed census open on the Ancestry website. The digitized fields are Name, Gender, Marital Status, Age, Estimated Birth Year, Residence Date, Home in 1931, Relation to Head,
Enumeration District and Sub-District. You can add fields for others in the household to the search form.

Having exhaused a very very short list of my relatives that were easily found I turned to the survivors of those who served with the Ottawa Company of Sharpshooters in 1885 believed to be living in Canada. None of the commissioned officers were alive. Of seven surviving NCOs one was clearly identified in the census, one was identified despite an ambiguity in the handwriting, candidates for two others were found, and three are not yet located.

Ancestry has done an amazing job in getting the name indexing completed in 9 days. The indexing will continue to improve and expand. Based on my limited testing you have a good chance of finding a person through the search. Your research skills will still come into play in resolving issues like name changes and handwriting ambiguity.

Please share your experience in the comments.

In correspondance with Ancestry I received the following information.

“The handwriting recognition model is set to try to detect to as close as possible the “as it appears” value on the image. In the example you have noted, a non-closed ‘g’ may in fact be captured as a y. While statistically speaking ‘Rogers’ would likely be a more common surname than ‘Royers’, what we have found is that if we overemphasize a check against statistically common names in our current AI data set, we run into the opposite issue where the less common results may be over written to a higher degree. As we continue to expand the locations, languages, and time periods of material that we process, however, we feel the AI will continue to learn and recognize some of the linguistic nuances that come with processing a large data set such as found in a national census, and will be better at determining what may be “sloppy penmanship” vs a valid, but less common name.”

I added

Interestingly, while there are numerous examples of Rogers being interpreted at Royers, there are no cases of George appearing as Georye, either as first or last name.

2 Replies to “Ancestry 1931 Canadian Census Name Indexed”

  1. I looked for my husband’s 2nd great-grandmother, Maria (Robinson) Eckford… Initial search using Maria Eckford, b 1850, in Vancouver gave me no results. Taking away her surname and adding widowed was a failure too. Then I specified North Vancouver (we didn’t know for sure where she was, given she wasn’t with her son, as expected) and the result Maria Scafford in North Van showed up. I clicked through, and sure enough, the AI had misread the Eck as Sca…I made the correction. We didn’t realize she lived separately from her family as she’d been with her son since her immigration in about 1908. I will have to now search directories to see when she first pops up separately.

  2. I noticed that the name of some of my French ancestors were often indexed incorrectly. Also found quite a few errors with the ages listed…

Leave a Reply

Your email address will not be published. Required fields are marked *