The 1917 Census of Altoona

A study of how AI can assist historical research

INTRODUCTION

In a previous blog post titled, “1917: Altoona officially becomes a town (Part 2 in a series on Waker Hale),” we explored the loss and reincorporation of Altoona as a town. Included in that post was a section pertaining to the census that was taken as part of the reincorporation process. While the Southern Democrat featured an article naming the total number of inhabitants, the oldest citizen, the youngest citizen, and other data, the census itself was not found during my research for that article.

Shortly after I made that blog post, someone sent me a picture and an accompanying message saying that the Etowah County courthouse had recently put together a display case showing some of the historical documents in its collection. They highlighted that some of these documents featured Altoona’s incorporation. When I viewed the photo, one thing that stuck out to me was the fact that a census was included in the Altoona documents. The next time I was at the courthouse, I made sure to check out the display case and saw that it was the 1917 census. It took me a couple of years to reach out, but I finally contacted Etowah County Probate Judge Scott Hassell, who was very gracious and accommodating in helping me make copies of this census. He personally removed the documents from the display case and allowed me to follow him to another room where the county has an overhead scanner for scanning these documents. Judge Hassell personally scanned them and saved them on a USB drive for me. I would be remiss if I did not thank him for the effort he made in bringing this census and accompanying research to light.

I would also like to take a moment to talk about the monumental task of digitizing all of the probate records that began under the previous judge, Bobby Junkins. While not the original goal of the project, this digitization effort has revolutionized historical research in Etowah County. Previously, when you were researching someone, a trip to the courthouse would mean accessing each record book’s index individually. As an example, if you were researching John Doe, you would have to examine the marriage book index to find his marriage record, the estate book or probate court minute index to find information on his estate, or the deed book index specific to his lifetime to find his property information. This involved going through handwritten alphabetical lists until you found John Doe’s name. When you finally found the book and page related to John Doe, you would have to consult the actual record book with the book letter and page number to find the original document with the information. With the digitization and indexing of these different record books, you can now search John Doe on a computer and pull up the scanned copy of each record relating to him. This has both simplified and expanded the potential for historical research within the county records. I have personally solved several dead ends because of this digitization project. It is my hope that one day all of our local counties will follow suit.

BACKGROUND ON THE 1917 CENSUS

As I previously mentioned, Altoona was required to take a census as part of the incorporation process. Under the current Alabama incorporation procedure, the proposed community first has to meet the basic statutory requirement of having not less than 300 people and must petition the county probate judge for incorporation. The petition process also requires a proposed name, an accurate plat, a boundary description, and proof related to residency and elector qualifications. The actual census, or enumeration, comes after the incorporation election, not before the petition. Alabama Code § 11-41-4 says that if a majority of votes are cast for “corporation,” the probate judge must cause an enumeration of the inhabitants within the proposed territory to be made. Probate records reveal that the official results of the incorporation petition were entered into the record on August 2, 1917.[1] At that time, Probate Judge Herzberg ordered that an enumeration of the inhabitants of Altoona be conducted. He named J.A. Webb and G.W. Eckles as the census takers.

Interestingly, neither Webb nor Eckles played any role within the town government, nor were they business owners. In fact, for all intents and purposes, the two census takers were common, everyday men, with both being employed as miners. George Washington Eckles was born to William James and Etta Eaton Eckles on December 25, 1878, in Gainesboro, Tennessee.[2] His father worked as a part-time farm hand and coal miner. By the time he was 21, George was working as a railroad section hand in Grundy, Tennessee.[3] By 1910, Eckles was residing in Altoona, where he appeared on the census as a miner living in the boarding house ran by Anne Choelia on Highland Street.[4] In 1918, he was shown as a mine foreman.[5] By 1920, he had relocated to Tuscaloosa County, and by 1930 he was in Acmar, Alabama by 1930. Although he was listed as a miner in both locations, family history states that Eckles became a successful grocer during his time in Acmar.[6] Eckles passed away at age 64 on October 6, 1943, and was buried at Cedar Grove Cemetery in Leeds.[7]

Jesse A. Webb was born in William and Frances Carr Webb in Cherokee County on May 27, 1890.[8] Webb’s formative years are relatively unknown. His parents were lifelong residents of Cherokee County and farmers. Jesse first appears in Etowah County records on January 13, 1913, when he married Bella Gash.[9] Gash was the daughter of Gorge Gash, a miner turned hotel owner who was appointed as an Altoona councilman on September 2, 1911.[10] The couple welcomed a daughter, Marguerite Webb, on August 3, 1915. While Webb’s employment at the time of the Altoona census is unknown, the following year he was employed as a policeman by the Tennessee Coal, Iron & Railroad Company in Ensley.[11] By 1920, he was working in the Altoona mines as a weighman and living in the household of his father-in-law.[12] Webb passed away in a Birmingham Hospital on February 17, 1925, and was buried in Elmwood Cemetery.[13]

Eckles and Webb completed the census on August 7, 1917, and were sworn before Justice of the Peace T. B. Wagnon. At that point, it was entered into the record of the probate court, and the remainder of the incorporation process continued normally. This census remained in the possession of the probate court and was relatively unknown until being put on display during the summer of 2024.

DIGITIZATION, ARTIFICIAL INTELLIGENCE AND THE TRANSCRIPTION PROCESS

I want to touch on the process that Judge Hassell used when scanning the original census. The county has a CZUR book scanner. Unlike a traditional flatbed or feed scanner, the book scanner sits on a desktop and uses a camera to photograph each individual page. Scanning the census was as simple as placing each page under the camera, which automatically detected the document and took a picture. The software automatically cropped the image to the edges of the document and placed it into a PDF.

The census was not a typed document. Instead, it was handwritten in cursive. When I began the transcription process, I created a Word document and made a table. The top of the table contained the page number, with the next row being a series of columns: surname, first name, age, sex, and race. The next row began the census itself, starting with surnames beginning with A. I split my screen into two different windows. The left window contained the Word document with the table, while the right window contained the PDF copy of the census. I would manually read and copy each row of the census into the Word document. This process was very time-consuming and frustrating. I transcribed all the “A” surnames this way until I began to wonder if ChatGPT could read cursive. I did a quick Google search and found that results utilizing ChatGPT for handwriting transcription were surprisingly positive. I uploaded the first page of the B surnames with the simple prompt, “Can you make this into plaintext in a cleaned-up table format?” The results were surprisingly good, and I quickly added another prompt asking it to export the table into a Word DOCX format.

I opened the results in Word and compared them to the original document. ChatGPT did a great job, but there were small errors. The most notable included confusion between the numbers two and three, between M and N, and some obvious misspellings of names. Despite this, utilizing ChatGPT to transcribe the handwritten census saved many hours of work. I quickly uploaded the rest of the document and had it transcribe the remaining twenty-something pages. After it finished, I copied the transcription into the original Word document. At that point, my workflow remained unchanged. The transcription in Microsoft Word was in the left pane, and the original handwritten census was in the right pane. I went line by line and checked for errors in transcription and formatting.

When I finished checking and correcting the entire census, I fed it back into ChatGPT to correct the race column. In the original census, individuals were either marked with a W for White or a C for Colored. In many cases, if several individuals were the same race, the census taker started marking their race with an X. I asked ChatGPT to replace each X with the previous letter, either W or C. I also asked it to fully spell out White or Colored. At this point, it generated a separate DOCX file, which I compared to my original to make sure that ChatGPT did not change any other lines or items. At this point, I felt the census was complete, and the entire document had been transcribed.

This is not to say that the transcription is without errors. Errors may remain from the AI-generated transcription, from the original census takers, and from my own work in the transcription. In many cases, it is obvious that the original census takers relied on phonetic spelling. The most obvious case of this is the Lindsey family, in which the original census taker spelled the surname Linsey. There are also numerous cases where the AI transcription was imperfect. When it was uncertain about a name, it would put a question mark beside it. However, as I went through the line-by-line check, I noticed several names that it got completely wrong but did not question at all. An example of this is the name Anna or Anne, which the AI handled poorly. Likewise, I am sure there are numerous cases where neither ChatGPT nor I could decipher a name.

In the end, since I was going to release the DOCX file along with this blog post, I decided to treat it as a living document and not as a true transcription of the original census. So, in the case of the Lindsay family, I corrected Linsey to Lindsay. Where Anna or Anne was obviously incorrect, I corrected it. I hope that, over time, more corrections will be made and that, in the end, this census transcription will be a useful document for future historians and genealogists, while the original handwritten census will remain the true original document.

With that in mind, I present the transcribed 1917 census of Altoona for use, correction, and reproduction: (PDF COPY AVAILABLE HERE)

State of Alabama

Etowah County

Personally before me T.N. Wagnon a Justice of the Peace in and for said County, comes J.A. Webb, and G.W. Eckles, who being duly sworn testifies and affirms that the above is a true and correct census of the population of the Town of Altoona, Alabama, to the best of their knowledge.

August 7, 1917

T.B. Wagnon, J.A. Webb, and G.W. Eckles (signatures)

[1] 10 August 1916, Birmingham Post-Herald, P.2.

[2] Eckles, Harry Edward. Eckles and Eckels Families in the United States: A Brief History of the Present Eckles and Eckels Families and Their An-Cestors, Together With Some Notes on the Origin of the Surname and Some Early General History, P. 149.

[3] Year: 1900; Census Place: Civil District 13, Grundy, Tennessee; Roll: 1573; Page: 6; Enumeration District: 0039.

[4] Year: 1910; Census Place: Altoona, Etowah, Alabama; Roll: T624_13; Page: 6b; Enumeration District: 0079; FHL microfilm: 1374026.

[5] U.S., World War I Draft Registration Cards, 1917-1918, Registration State: Alabama; Registration County: Etowah County, registration card number 4460.

[6] Ibid., #2,

[7] Find a Grave, database and images (https://www.findagrave.com/memorial/37191316/george_w-eckles: accessed June 23, 2026), memorial page for George W. Eckles (25 Dec 1879–6 Oct 1943), Find a Grave Memorial ID 37191316.

[8] Shoop, Thorpe Family Tree. Ancestry user MThrope55, https://www.ancestry.com/family-tree/person/tree/27096394/person/26894890756/facts.

[9] Etowah County Marriage Record Book 5, Page 348.

[10] Altoona Council Minute Book 1908-1910, September 2, 1911, council meeting. Year: 1920; Census Place: Altoona, Etowah, Alabama; Roll: T625_15; Page: 8B; Enumeration District: 118.

[11] U.S., World War I Draft Registration Cards, 1917-1918, Registration State: Alabama; Registration County: Etowah County, registration card number 109.

[12] Year: 1920; Census Place: Altoona, Etowah, Alabama; Roll: T625_15; Page: 9A; Enumeration District: 118

[13] 18 February 1925, “J.A. Webb.” The Birmingham News, P. 2.